Patient Dose in Brain CT Perfusion - Semantic Scholar...Computed Tomography Perfusion exams but also...
Transcript of Patient Dose in Brain CT Perfusion - Semantic Scholar...Computed Tomography Perfusion exams but also...
INTERDEPARTMENTAL PROGRAMME OF
POSTGRADUATE STUDIES IN
MEDICAL PHYSICS
Patient Dose in Brain CT Perfusion
DIMITRIOS N. GEORGAKOPOULOS
M.Sc. Thesis
October 2017 Patras
ΔΙΑΤΜΗΜΑΤΙΚΟ ΠΡΟΓΡΑΜΜΑ
ΜΕΤΑΠΤΥΧΙΑΚΩΝ ΣΠΟΥΔΩΝ ΣΤΗΝ
ΙΑΤΡΙΚΗ ΦΥΣΙΚΗ
Δόση ασθενούς στην
Αξονική Τομογραφία
αιμάτωσης εγκεφάλου
ΔΗΜΗΤΡΙΟΣ Ν. ΓΕΩΡΓΑΚΟΠΟΥΛΟΣ
ΜΕΤΑΠΤΥΧΙΑΚΗ ΔΙΠΛΩΜΑΤΙΚΗ ΕΡΓΑΣΙΑ
ΟΚΤΩΒΡΙΟΣ 2017 ΠΑΤΡΑ
THREE MEMBER EXAMINATION COMMITTEE
Panayiotakis George, Professor of Medical Physics (Supervisor)
Kalogeropoulou Christina, Associate Professor of Radiology
Zampakis Petros, Assistant Professor of Radiology
ACKNOWLEDGEMENTS
I would like to thank my thesis supervisor, Professor George Panayiotakis for giving
me the opportunity to realize this project and for his continuous support and guidance
during this Master Thesis.
I am also indebted to the Medical Radiation Physicist from the University Hospital
of Patras, Dr. Gerasimos Messaris for his valuable scientific support and co-operation
during this research. Without his knowledge and his advices this thesis would not have
been possible.
My sincerest appreciation to the Associate Professor Christina Kalogeropoulou and
the Assistant Professor P. Zampakis from the Department of Radiology, for giving me
the opportunity not only to have access to the patient’s data regarding the specific
Computed Tomography Perfusion exams but also for their valuable discussions
regarding the medical aspects of the brain perfusion imaging procedures.
Last but not least, I would like to thank the radiation technologists Ms. Vicky
Vavatsikou and Mr. Epameinondas Ntzanis for their valuable help during the collection
of the specific exams.
[i]
CONTENTS
ABSTRACT…………………………………………………………………………...1
ΠΕΡΙΛΗΨΗ…………………………………………………………………………...2
GENERAL PART………………………………………………………………...3
INTRODUCTION……………………………………………………………………..3
Chapter 1. Principles of Computed Tomography……………………………..4
1.1. X-ray projection, attenuation and acquisition of transmission profiles………4
1.2. Hounsfield units……………………………………………………………...6
1.3. The CT Imaging System……………………………………………………..7
1.4. Gantry and Table…………………………………………………………….8
1.5. The X-ray tube and generator………………………………………………..9
1.6. Collimation and filtration………………………………………………….....9
1.7. Detectors……………………………………………………………………12
1.8. Image Reconstruction and Processing………………………………………15
1.9. Object space, image space and Radon space……………………………….16
1.10. Filtered back projection and other reconstructions………………………...17
1.11. Axial CT scan………………………………………………………………21
1.12. Helical CT scan…………………………………………………………….22
1.13. MultiDetector CT (MDCT) scan…………………………………………...23
1.14. Contrast Enhanced CT…………………………………………………..….24
1.15. Special Applications………………………………………………………..24
1.16. CT image quality…………………………………………………………...25
1.17. Effect of acquisition and reconstruction parameters on image quality…….28
1.18. Artifacts…………………………………………………………………….30
Chapter 2. CT dose descriptors…………………………………………………32
2.1. Radiation Dose Measures: General Definitions………………………………..32
2.2. Radiation Dose Measures: CT Specific………………………………………..34
2.3. Factors That Influence Radiation Dose from CT……………………………….42
2.4. Effects of Object (and Patient) Size……………………………………………47
2.5. Other Options for Reducing Scan Dose………………………………………..48
2.6. Indirect Effects………………………………………………………………....48
2.7. Methods to Reduce Patient Dose………………………………………………49
[ii]
SPECIFIC PART………………………………………………………………...51
Chapter 3. CT imaging modalities……………………………………………...51
3.1. Theoretic Basis………………………………………………………………...51
3.2. Unenhanced Brain CT (UN-CT) ……………………………………………...52
3.2.1. Detection of hemorrhage or stroke mimickers………………………...….52
3.2.2. Distinction between Ischemia and Hemorrhage…………………..………53
3.2.3. Detection of ischemic signs of established infarction…………………….53
3.2.4. Subtle early ischemic signs…………………………………………….….54
3.2.5. Detection of early acute ischemic stroke on nonenhanced CT……………54
3.2.6. Sensitivity for depiction of subtle early signs of infarction and ischemia...55
3.3. Brain CT Angiography (CTA) ………………………………………………..56
3.3.1. Delineation of the presence and site of vascular occlusion……………….56
3.3.2. Depiction of arterial dissection……………………………………………56
3.3.3. Grading of collateral blood flow…………………………………………..59
3.3.4. Characterization of atherosclerotic disease………………………………..64
3.4. Brain CT Perfusion (CTP)……………………………………………………..65
3.4.1. CT Perfusion general principles…………………………………..………65
3.4.2. Quantitative Analysis in CT Perfusion……………………………………68
3.4.3. Qualitative Analysis in CT Perfusion……………………………………..69
3.4.4. Calculation of the specific parameters: MTT, CBV, CBF………………..71
3.4.5. Multimodal CT Evaluation Aim…………………………………………..83
Chapter 4. CT Perfusion as a diagnostic tool…………………………………84
4.1. CT Perfusion in acute stroke……………………………………………………84
4.2. Evaluation and Anatomical Region…………………………………………….84
4.3. Associated Ionizing Radiation…………………………………………………84
4.4. Aim of the study………………………………………………………………..85
Chapter 5. CT perfusion in clinical practice…………………………………..85
5.1. Patient data and utilized CT system……………………………………………85
5.2. Protocol and contrast agent parameters………………………………………..85
5.3. Dose Report data……………………………………………………………….86
5.4. Conversion Factor for the estimation of the Effective Dose……………………86
[iii]
Chapter 6. Dosimetry Results……………………………………………………88
6.1. CTDIvol for the comprehensive prescription protocol…………………………..88
6.2. DLP for the comprehensive prescription protocol……………………………..88
6.3. Effective Dose for the comprehensive prescription protocol…………………..89
Chapter 7. Discussion and Conclusion…………………………………………91
7.1. Comparison of CTDIvol with the literature……………………………………..91
7.2. Comparison of DLP with the literature…………………………………………91
7.3. Comparison of Effective Dose with the literature………………………………92
7.4. Comparison of Total Effective Dose with the literature……………………….94
7.5. Comparison of the Protocol Effective Dose with Background Radiation……..94
7.6. Advantages and Limitations of the study………………………………………95
7.7. Future work………………………………………………………………….....95
7.8. Conclusion……………………………………………………………………..96
APPENDIX………………………………………………………………………...99
REFERENCES…………………………………………………………………...101
1
ABSTRACT
Background and purpose. - Brain CT Perfusion (CTP) is an X-ray imaging modality
for the assessment of cerebrovascular disorders. The aim of this study is the evaluation
of the radiation dose to patients during a comprehensive brain CT prescription protocol
(CPP) consisting of an unenhanced brain CT, a brain CT Angiography and a CTP scan.
Materials and Methods. - Fifteen patients were studied using an 80-slice CT system,
with an iterative reconstruction algorithm. The volume Computed Tomography Dose
Index (CTDIvol) and Dose Length Product (DLP) were recorded from the dose report
of the system. The calculation of Effective Dose (ED) was accomplished using the DLP
values.
Results. - For the CTP examinations, the CTDIvol ranged from 116.0 to 134.8 mGy,
with the mean value 119.8 mGy. The DLP ranged from 463.9 to 539.2 mGy∙cm, with
the mean value 478.9 mGy∙cm. For the CPP, the total ED ranged from 3.31 to 4.88
mSv, with the mean value 4.34 mSv.
Conclusions. - The radiation dose to the patients from the 80-slice CT system of our
study is lower not only than that of 64-slice CT systems but also lower than that of
128,192,256 and 320-slice CT systems, reported in corresponding studies.
2
ΠΕΡΙΛΗΨΗ
Το CT Perfusion εγκεφάλου (CTP) είναι μια μορφή απεικόνισης με ακτίνες Χ για την
αξιολόγηση εγκεφαλοαγγειακών διαταραχών. Ο σκοπός αυτής της μελέτης είναι η
αξιολόγηση της δόσης ακτινοβολίας σε ασθενείς κατά τη διάρκεια ενός συνεκτικού
πρωτοκόλλου αξονικής τομογραφίας εγκεφάλου που αποτελείται από μια αξονική
τομογραφία εγκεφάλου (UN-CT), μια αξονική αγγειογραφία εγκεφάλου (CTA) και μια
σάρωση αξονικής τομογραφίας αιμάτωσης εγκεφάλου (CTP).
Μελετήθηκαν δεκαπέντε ασθενείς χρησιμοποιώντας ένα σύστημα αξονικής
τομογραφίας 80 τομών, με έναν επαναληπτικό αλγόριθμο ανακατασκευής. Ο
ογκομετρικός δείκτης δόσης υπολογιστικής τομογραφίας (CTDIvol) και το γινόμενο
δόσης-μήκους σάρωσης (DLP) καταγράφηκαν από την αναφορά δόσης του
συστήματος. Ο υπολογισμός της ενεργού δόσης (ED) πραγματοποιήθηκε
χρησιμοποιώντας τις τιμές του DLP.
Για τις εξετάσεις CTP, το CTDIvol κυμάνθηκε από 116.0 έως 134.8 mGy, με μέση
τιμή 119.8 mGy. Το DLP κυμάνθηκε από 463.9 έως 539.2 mGy ∙ cm, με τη μέση τιμή
478.9 mGy ∙ cm. Για το συνεκτικό πρωτόκολλο, η συνολική ED κυμάνθηκε από 3.31
έως 4.88 mSv, με τη μέση τιμή 4.34 mSv.
Η δόση ακτινοβολίας στους ασθενείς από το σύστημα αξονικής τομογραφίας των 80
τομών της μελέτης μας είναι χαμηλότερη όχι μόνο από εκείνη των συστημάτων
αξονικής τομογραφίας 64 τομών αλλά χαμηλότερη και από αυτή των συστημάτων
αξονικής τομογραφίας 128,192,256 και 320 τομών, που αναφέρονται σε αντίστοιχες
μελέτες.
3
GENERAL PART
INTRODUCTION
Computed Tomography (CT) scanner technology has been developed significantly
since the first CT scanner was constructed in early 1970s. Initial CT scanners were
single slice axial, but technological development has seen the introduction of helical
and multi-slice models. Modern scanners are capable of imaging simultaneously 64,
128 or even 320 parallel slices in one rotation. Beam width has increased significantly
from a standard of 10 mm to current beam widths of up to 160 mm. The use of CT has
been increasing rapidly; there have been 12-fold and 20-fold increases in CT in
European countries and the United States over the last 20 years. Moreover, CT is a high
radiation dose examination and makes the largest contribution to the patient radiation
dose from medical exposures. CT now accounts for 50%, 68% and 70 % of the
collective dose in European countries, the United Kingdom and the United States,
respectively. Because the use of CT has been increasing rapidly, there has been growing
concern about potential health effects from the high doses that can be delivered, and
patient dose from CT examinations has become a cause for concern among radiological
professionals.
The standard method for CT dosimetry measurement has been the CT dose index
(CTDI). This is designed to measure the output for a single CT slice or a limited number
of slices, but is also used for measurements inside phantoms simulating parts of the
body for the purpose of patient dose assessment. Scans with an axial slice include most
of the radiation within the length of the standard 100 mm pencil chamber used for the
measurement.
The aim of this study is the evaluation of the radiation dose to patients during a
comprehensive brain CT prescription protocol (CPP) consisting of an unenhanced brain
CT, a brain CT Angiography and a CTP scan, using the CTDIvol and DLP values from
an 80-slice CT system at the University Hospital of Patras.
4
Chapter 1. Principles of Computed Tomography
1.1. X-ray projection, attenuation and acquisition of transmission profiles
The process of CT image acquisition involves the measurement of X-ray transmission
profiles through a patient for a large number of views. A profile from each view is
achieved primarily by using a detector arc generally consisting of 800–900 detector
elements (dels), referred to as a detector row. By rotation of the X-ray tube and detector
row around the patient, a large number of views can be obtained. The use of tens or
even hundreds of detector rows aligned along the axis of rotation allows even more
rapid acquisition (Fig. 1). The acquired transmission profiles are used to reconstruct the
CT image, composed of a matrix of picture elements (pixels).
Fig. 1. CT image acquisition showing the transmission of X rays through the patient by using
a detector row (a), with rotation of the X ray tube and detector (b) and by multiple detector (c).
The values that are assigned to the pixels in a CT image are associated with the
attenuation of the corresponding tissue, or, more specifically, to their linear attenuation
coefficient μ (m-1). The linear attenuation coefficient depends on the composition of the
material, the density of the material and the photon energy, as seen in Beer’s law:
I(x) = Io ∙ e -μ∙x (1)
where I(x) is the intensity of the attenuated X-ray beam, Io the unattenuated X-ray
beam and x the thickness of the material. Note that Beer’s law only describes the
attenuation of the primary beam and does not take into account the intensity of scattered
radiation that is generated. For use in polyenergetic X ray beams, Beer’s law should
5
strictly be integrated over all photon energies in the X-ray spectrum. however, in the
back projection methodologies developed for CT reconstruction algorithms, this is
generally not implemented; instead, typically, a pragmatic solution is to assume where
Beer’s law can be applied using one value representing the average photon energy of
the X ray spectrum. This assumption causes inaccuracies in the reconstruction and leads
to the beam hardening artefact.
As an X ray beam is transmitted through the patient, different tissues are encountered
with different linear attenuation coefficients. If the pathway through the patient ranges
from 0 to d, then the intensity of the attenuated X-ray beam, transmitted a distance d,
can be expressed as:
I(d) = Io ∙ 𝑒− ∫ 𝜇(𝑥)𝑑𝑥
𝑑
0 (2)
Since a CT image is composed of a matrix of pixels, the scanned patient can also be
regarded as being made up of a matrix of different linear attenuation coefficient volume
elements (voxels). Figure 2 shows a simplified 4 × 4 matrix representing the
measurement of transmission along one line. For such a discretization, the equation for
the attenuation can be expressed as:
I(d) = Io ∙ 𝑒− ∑ (𝜇𝑖𝛥𝑥)𝑖=4𝑖=1 (3)
Fig. 2. The principle of attenuation of an X ray beam in a simplified 4 × 4 matrix. Each element
in the matrix can, in principle, have a different value of the associated linear attenuation
coefficient.
6
From the above, it can be seen that the basic data needed for CT are the intensities of
the attenuated and unattenuated X-ray beams, respectively I(d) and Io, and that these
can be measured. Image reconstruction techniques can then be applied to derive the
matrix of linear attenuation coefficients, which is the basis of the CT image.
1.2. Hounsfield units
In the CT image, the matrix of reconstructed linear attenuation coefficients (μmaterial)
is transformed into a corresponding matrix of Hounsfield units (HUmaterial), where the
HU scale is expressed relative to the linear attenuation coefficient of water at room
temperature (μwater):
HUmaterial = μmaterial - μwater ∙ 1000 (4)
μwater
It can be seen that HUwater = 0 when μmaterial = μwater, HUair = –1000 when μmaterial = 0
and HU = 1 is associated with 0.1% of the linear attenuation coefficient of water. Table
1 shows typical values for body tissues. From the definition of the HU, it follows that
for all substances except water and air, variations of the HU values occur when they are
determined at different tube voltages. The reason is that, as a function of photon energy,
different substances exhibit a non-linear relationship of their linear attenuation
coefficient relative to that of water. This effect is most notable for substances that have
a relatively high effective atomic number, such as contrast enhanced blood and bone.
TABLE 1. TYPICAL HU VALUES AND RANGES OF VALUES FOR DIFFERENT
TISSUES AND MATERIALS
7
The minimum bit depth that should be assigned to a pixel is 12, enabling the creation
of a Hounsfield scale that ranges from –1024 HU to +3071 HU, thus covering most
clinically relevant tissues. A wider Hounsfield scale with a bit depth of 14 is useful for
extending the HU scale upwards to +15 359 HU, thus making it compatible with
materials that have a high density and a high linear attenuation coefficient.
CT images are usually visualized on a monitor using an eight bit greyscale offering
only 256 grey values. Each pixel HU value then has to undergo a linear mapping to a
‘window’ 8 bit value. The window width defines the range of HU’s that is represented
by the mapped values (ranging from white to black) and the window level defines the
central HU value within the selected window width. Optimal visualization of the tissues
of interest in the image can only be achieved by selecting the most appropriate window
width and window level. Consequently, different settings of the window width and
window level are used to visualize soft tissue, lung tissue or bone. The greyscale, as
defined by window level and window width, is adapted to the diagnostic task and is
thus dependent on the clinical question.
In clinical practice, considerable deviations between the expected and the observed
HU values may occur. Causes for such inaccuracies may, for example, be the
dependence of the HU value on the reconstruction filter, the size of the scanned field
of view (FOV), and the position within the scanned FOV. In addition, image artifacts
may have an effect on the accuracy of the HU values. When performing longitudinal
clinical studies, one should take into account that, even for the same scanner, the HU
values for a given tissue type may vary with time. In multicenter studies that involve
different CT scanners, there may also be significant variations in the observed HU
values. Therefore, quantitative imaging in CT requires special attention and often
additional calibrations of the CT scanner.
1.3. The CT Imaging System
The development of multidetector row CT (MDCT) and multisource CT had already
been described in a US patent from 1980. The patent describes what the authors call “a
multiple purpose high speed tomographic X ray scanner”. In the acquisition technique
of helical CT, the patent states that “the apparatus enables helical scanning to be
effected by the continuous transportation of the table couch”. The helix is the pathway
of the continuously rotating X ray source seen from the perspective of the patient.
8
Fig. 3. Concepts of multisource and MDCT scanning (left) and of helical CT (right).
Currently, most scanners are helical MDCT scanners, but the technologies of dual
source and volumetric CT scanning have been implemented on a wide scale.
1.4. Gantry and Table
The gantry contains all the system components that are required to record
transmission profiles of the patient. Since transmission profiles have to be recorded at
different angles, these components are mounted on a support within the gantry that can
be rotated. The X-ray tube with high voltage generator and tube cooling system, the
collimator, the beam shaping filters, the detector arc and the data acquisition system are
all mounted on this support. The engineering of these components is complex, since
they need to be able to withstand the strong centrifugal force that occurs during the fast
rotation of the gantry. Forces of several tens of G’s arise for rotation times of the order
of 0.25 s.
Electrical power is generally supplied to the rotating gantry by means of slip ring
contacts. Recorded projection profiles are generally transmitted from the gantry to a
computer by means of wireless communication technologies.
The design and engineering of the table, as with the gantry, are critical to allowing
accurate acquisition of data at high rotational speeds. The table must also be able to
withstand heavy weights without bending. The position of the patient on the table can
be head first or feet first, and supine or prone; this position is usually recorded with the
scan data.
9
TABLE 2. OVERVIEW OF DIFFERENT TYPES OF CT TECHNOLOGY
1.5. The X-ray tube and generator
Owing to the high X-ray flux required for CT, the X-ray tube uses a tungsten anode
designed to withstand and dissipate high heat loads. With long continuous acquisition
cycles, a forced cooling system using oil or water circulated through a heat exchanger
is often used.
1.6. Collimation and filtration
The X-ray beam should be collimated to the desired dimensions. The beam
collimation for defining the thickness of the slice to be imaged is made in the first
instance close to the X-ray source (primary collimation). The shape of the dose profile
is determined by the aperture of the collimator, its distance from the focal spot, and the
size and shape (i.e. the intensity distribution) of the focal spot. Due to the narrow width
10
of collimation, penumbral effects occur. These effects become more and more
pronounced as collimation is further narrowed.
In addition, there is a secondary collimation close to the detector (‘post-patient
collimation’) that primarily serves to remove scattered radiation. On some single-slice
and dual-slice scanners this secondary collimation is further narrowed in order to
improve the shape of the slice profile (‘restrictive post-patient collimation’, Fig. 4 a,b).
For multi-slice scanners with more than two detector rows, the primary collimation
must necessarily be made wider than N times the selected slice collimation in order to
avoid (or at least to reduce) penumbral effects in the outer portions of the detector array
(Fig. 4c). In both cases, the dose profile is wider than the slice profile or the nominal
beam width, and the patient is exposed to a larger extent (‘overbeaming’), as becomes
obvious from normalized CTDI values that increase with reduced beam width.
Overbeaming can be expressed by a single parameter, the ‘overbeaming parameter’
dz, that is equal to the combined width of the portion of the dose profile that is not used
for detection (fig. 4c). Overbeaming itself, i.e. the percentage increase in CTDI due to
the unused portion of the dose profile, is then given by :
ΔCTDI = (dz / NxT) ∙ 100 (5)
With multi-slice scanners, overbeaming effects have to be taken seriously, as MSCT
technology aims to provide improved resolution along the z-axis, which requires
reduced slice collimation. Overbeaming, i.e. the increase in CTDI that results from
beam width settings that are typical for each type of scanner is shown in Fig. 5 for a
number of scanners from different manufacturers. As indicated by the trend line,
overbeaming is most pronounced with quad-slice scanners and is diminished with an
increasing beam width NxT (N∙hcol) provided by scanners with more slices.
11
Fig.4. Dose profiles free-in-air with umbra (dark grey) and penumbra (light grey) portions for
a single-slice scanner (a.), a dual-slice scanner (b.), and a quad-slice scanner (c.). With single-
and dual-slice scanners, the width of the active detector rows is sufficient to capture the entire
dose profile, penumbra included (except for some scanners which employ restrictive post-
patient collimation). For scanners with four or more slices acquired simultaneously, penumbra
is excluded from detection to serve all detector channels equally well. The combined width of
the penumbra triangles at both sides is characterized by the overbeaming parameter dz.
Fig.5. Overbeaming, i.e. the percentage increase in CTDI, for single-slice (N=1), dual-slice
(N=2), quad-slice (N=4), 6 to 8-slice (N=6-8), 16-slice (N=16) and 32 to 40-slice (N=32-40)
scanners from different manufacturers (A to F) for the slice collimations hcol typically
employed. The red trend line indicates that overbeaming is most pronounced with quad-slice
scanners in practice and is diminished with an increasing beam width N·hcol.
The present generation of scanners typically employs a beam filtration for the X-ray
tube assembly of between 1 and 3 mm Al and an additional filtration (flat filter) of 0.1
mm Cu, giving a total beam filtration of between 5 to 6 mm Al. The use of additional
filtration impairs primary contrast and increases noise due to reduced beam intensity
per mAs as experienced by the detectors. Without compensating for these adverse
effects (e.g. by increasing tube current-time product), the contrast-to-noise ratio, which
affects the detectability of small or low-contrast details, is reduced.
The beam width in the longitudinal axis is generally small; therefore, the collimated
X-ray beam is often referred to as a fan beam. In the plane perpendicular to the table
motion, also known as the x–y or axial plane, the beam is shaped to reduce the dynamic
range of the signal that is recorded by the detectors.
Beam shaping (bowtie) filters modify the spatial distribution of radiation emitted
within the fan beam and are used to achieve the desired gradient, with one of a number
of mounted bowtie filters moved into the X-ray beam during acquisition. The purpose
of this kind of filter (which is characterized by increasing thickness towards its outer
12
edges) is to adapt the beam intensity to match the reduced attenuation of objects in the
outer portions of the fan beam.
In order to provide attenuating properties that are almost tissue equivalent, beam
shapers should be made from materials containing only elements with a low atomic
number Z. However, this is not always the case in practice. Beam shapers preferentially
affect the dose in the outer portions of an object, thereby reducing the peripheral CTDIp
values. But as the dose at the centre is mainly caused by scattered radiation from the
periphery of the object, the central CTDIc value is also somewhat reduced. The ratio of
dose at the periphery to the dose at the centre therefore decreases, making the dose
distribution inside an object more homogeneous and so improving the uniformity of
noise in the image. Contrary to the flat filter, the beam shaper has a much greater impact
on the dose properties of a scanner.
1.7. Detectors
The essential physical characteristics of CT detectors are a good detection efficiency
and a fast response with little afterglow. Currently, solid state detectors are used, as
they have a detection efficiency close to 100% compared with high pressure, xenon
filled ionization chambers that were used previously and that had a detection efficiency
of about 70%. Solid state detectors are generally scintilators, meaning that the X-rays
interacting with the detector generate light. This light is converted to an electrical
signal, by photodiodes that are attached to the back of the scintillator, which should
have good transparency to ensure optimal detection. Typically, an antiscatter grid is
mounted at the front of the detector, which consists of small strips of highly attenuating
material (e.g. tungsten) aligned along the longitudinal (z) axis of the CT scanner,
forming a 1-D antiscatter grid.
A detector row consists of thousands of dels that are separated by septa designed to
prevent light generated in one del from being detected by neighbouring dels. These
septa and the strips of the antiscatter grid should be as small as possible since they
reduce the effective area of the detector and thus reduce the detection of X-rays. Figure
4 shows detector modules for a 4, 16, 64 and 320 slice CT scanner. The complete CT
detector is composed of many detector modules that are mounted next to each other
along an arc. CT detectors are curved in the axial (x–y) plane and rectangular along the
longitudinal (z) axis. While most dels are used to measure transmission profile data (the
13
attenuated intensity I(d)), the dels outside the FOV are used to measure the unattenuated
intensity of the X ray beam (Io). Thus, the coefficient I(d)/Io from Eq. (2) can be easily
recorded.
Fig. 6. Detector modules for a 4, 16, 64 and 320 slice CT scanner (left). The complete
CT detector is composed of many detector modules (right) (Toshiba Medical Systems).
The smallest size of an object (d) within the patient that can be resolved in the
reconstructed image depends on the number and size of dels along the detector arc, the
size of the dels along the z axis, the number of angles for which projections are recorded
during the acquisition, and the focal spot size of the X-ray tube. The minimum number
of dels in a detector arc covering a specific FOV should be approximately 2∙FOV/d, to
resolve the object, d, in the reconstructed image. About 800 dels are required to achieve
a spatial resolution of 1 mm within a reconstructed image for a FOV of 400 mm. Spatial
resolution can be improved for an acquisition with a full 360º rotation by a slight
geometrical modification of the arrangement of the dels. By shifting the dels by a
distance equal to a quarter of their size, the theoretically achievable spatial resolution
becomes twice as good. Thus, a quarter detector shift is generally implemented in CT
scanners. As a rule of thumb, the number of required projection angles can be
approximated by the number of required dels. With the current detector rows of 800–
1000 dels, covering an FOV of 400 mm, a spatial resolution of better than one
millimeter can be achieved.
Figure 7 shows how coverage of MDCT scanners increased when more active
detector rows became available. A typical acquisition with a single detector row scanner
covered 5 mm. CT scanners with four active detector rows achieved a substantial
improvement of the longitudinal resolution. For example, by using four active detector
rows in a 4 × 1 mm acquisition configuration, the longitudinal spatial resolution
improved from 5 mm to 1.25 mm.
14
The single detectors in a multi-row, solid-state detector array are separated by narrow
strips (‘septa’) which are not sensitive to radiation and therefore do not contribute to
detector signal. Due to the large number of additional strips, these inactive zones result
in minor or major geometrical losses, depending on the design of the detector array. In
addition, further losses occur due to a decrease in sensitivity at the edges of each row
that results from cutting the scintillator crystal. In contrast to a single-row detector array
whose width can be larger than the maximum slice thickness (see fig. 8), the edges of
the rows in a multirow detector array are located inside the beam. Due to both these
effects - separating strips and decreased sensitivity - the net efficiency of a solid-state
detector array, which is typically 85% for single-slice scanners, is further decreased to
typically 70%.
In clinical practice, CT scanners with four active detector rows were primarily used
to enhance longitudinal resolution, which allowed 3-D visualization of the scanned
volume. The CT scanners with four active detector rows could also be used for
enhanced longitudinal coverage, for example, by selecting a 4 × 2 = 8 mm, or even a 4
× 4 = 16 mm coverage.
Fig. 7. Coverage of the MDCT scanners increased when more
active detector rows became available.
15
Fig. 8. MSCT scanner, with simultaneous scanning of four slices, compared with a conventional
single-slice scanner. Due to the additional septa between the detector rows, the geometric
efficiency of MSCT detector arrays is comparatively lower by 10 to 20%.
Enhanced longitudinal coverage would allow for shorter scan times but without the
benefit of improved longitudinal resolution. The CT scanners with 16 or 64 active
detector rows allowed for acquisitions in, for example, 16 × 0.5 = 8 mm and 64 × 0.5 =
32 mm configurations. These scanners provided excellent longitudinal spatial
resolution, high quality 3-D reconstructions and, at the same time, reduced scan times.
The MDCT scanners with up to 64 active detector rows do not provide coverage of
entire organs, and to cover the prescribed range, the scan is generally a helical
acquisition with multiple rotations. With the 320 detector row CT scanner, one single
rotation allows for coverage of 160 mm, enough for covering organs such as the brain
or the heart within one single rotation.
1.8. Image Reconstruction and Processing
In order to reconstruct a CT image, numerous measurements of the transmission of
X-rays through the patient are acquired. This information is the basis for reconstruction
of the CT image. Prior to image reconstruction, a logarithm of the measured data is
calculated. The logarithm of the (inverse) measured normalized transmission,
ln(Io/I(d)), yields a linear relationship with the products of μi ∙ Δx (Eqs 2,3).
Intuitively, one might consider that a simple back projection of measured
transmission profiles could be used for image reconstruction. This process is visualized
in fig. 9, which shows (a) the X-ray projection at a certain angle producing a
transmission profile, (b) the back projection of this profile distributes the measured
signal evenly over the area at the same angle as the projection, (c) on addition of the
back projections of the transmission profiles from all projection angles, it becomes clear
that the simple back projection process yields a strongly blurred image and (d) a more
accurate reconstruction can be obtained by filtering the profiles prior to back projection.
This is the method of filtered back projection, which is discussed in the following
sections, and is the standard technique used for image reconstruction in CT.
16
Fig. 9. A simple back projection yields a strongly blurred image. The contours of the chest
and lungs can still be recognized in the image.
1.9. Object space, image space and Radon space
In order to understand the technique of filtered back projection better, it is essential
to introduce three interrelated domains: (i) the object space (linear attenuation values),
(ii) the Radon space (projection values, this domain is also referred to as sinogram
space, in which case Cartesian coordinates are used) and (iii) the fourier space, which
can be derived from the object space by a 2-D Fourier Transform.
Figure 10 illustrates the interrelations between the three domains for one projection
angle with the transmission projection (b) at one specific projection angle; this
projection corresponds with one line in Radon space (c). A 1-D Fourier Transform of
the recorded line in the sinogram yields an angulated line in fourier space (d).
Fig. 10. Projection (b) recorded by the CT scanner for (a); one specific projection angle
corresponds to one line in Radon space (c) and a 1-D FT of the recorded line in the sinogram
yields one line in Fourier space (d) at the same angle.
The interrelationships between the three domains, object space, Radon space and
Fourier space, are illustrated in fig. 11. A 2-D Radon transform converts the object
space into Radon space. The 2-D Radon space is actually created during the CT scan:
projections are recorded and stored as raw data in 2-D Radon space.
17
As will be shown in the next section, the combination of 1-D FT’s of transmission
profiles at many angles allows creation of the Fourier space of the object space. One
could intuitively expect that an inverse 2-D FT of Fourier space would be used in CT
to reconstruct the object space. However, this does not yield the best result, since the
rebinning of the Fourier transformed angulated projections, and the associated
interpolations that are required to achieve a Fourier space in cartesian coordinates, are
prone to induce artifacts in the reconstructed images. A better technique for CT
reconstruction is to use a filtered back projection.
Fig. 11. The interrelationships between the three domains, object space, Radon space and
Fourier space. Note that multiple 1-D FTs of lines in the Radon space allow the creation of the
2-D Fourier space (the number of 1-D transforms is equal to the number of profiles registered).
1.10. Filtered back projection and other reconstructions
The mathematical operations that are required for a filtered back projection consist of
four steps, which are elaborated in the following paragraphs. First, a FT of Radon space
should be performed (requiring many 1-D FT’s). Then, a high pass filter should be
applied to each one of the 1-D FT’s. Next, an inverse FT should be applied to the high
pass filtered FT’s, in order to obtain a Radon space with modified projection profiles.
Finally, back projection of the filtered profiles yields the reconstruction of the measured
object. Figure 12 illustrates this by showing how successive filtered back projections at
different angles can be used to achieve a good reconstruction of the space domain. It
may be noted at this stage that (in accordance with the convolution theorem for FT’s)
the filter that is applied to the Fourier domain can be substituted by a direct convolution
of profiles in the Radon domain with an appropriate kernel.
18
Image space is generally represented on a regular grid. Let the 2-D image space be
defined as ƒ(x, y), where (x, y) are rectangular cartesian coordinates. A single 1-D
projection of the 2-D image space with equidistant and parallel rays yields one line in
Radon space, expressed as the projection p(t, θ), where t is the distance from the
projected X ray to the isocentre and θ is the projection angle (Fig. 13). The central slice
theorem, also referred to as the Fourier slice theorem, states that the FT of such a
parallel projection of image space at the projection angle θ yields one line in 2-D
Fourier space, F(u, v), angulated at the same angle θ (the 2-D Fourier space is
sometimes also referred to as k space).
Fig. 12. Successive filtered back projections can be used to achieve a good reconstruction of
the space domain. The images are associated with, respectively, 1, 2, 4, 8, 16, 32, 64, 256 and
1024 filtered back projections at different angles.
This can be demonstrated as follows. At the projection angle θ = 0, the projection p(x,0)
and the corresponding line in Radon space is described as:
p (x,0) = ∫ 𝑓(𝑥, 𝑦)𝑑𝑦+∞
−∞ (6)
The 1-D FT with respect to x, of the projection p(x, 0) at the projection angle θ = 0 is
given by:
P(u) = ∫ 𝑝(𝑥, 0)𝑒−𝑖2𝜋𝑢𝑥𝑑𝑥 = ∫+∞
−∞∫ 𝑓(𝑥, 𝑦)𝑒−𝑖2𝜋𝑢𝑥𝑑𝑥𝑑𝑦
+∞
−∞
+∞
−∞ (7)
and the 2-D ft F(u, v) of the 2-D image space ƒ(x, y) at v = 0 is:
19
F(u,v)│v=0 =∫+∞
−∞ ∫ 𝑓(𝑥, 𝑦)𝑒−𝑖2𝜋(𝑢𝑥+𝑣𝑦)𝑑𝑥𝑑𝑦+∞
−∞│v=0 =∫
+∞
−∞ ∫ 𝑓(𝑥, 𝑦)𝑒−𝑖2𝜋𝑢𝑥𝑑𝑥𝑑𝑦+∞
−∞ (8)
It thus becomes clear that the 1-D FT with respect to x for the projection angle θ = 0 is
equal to the 2-D FT F(u, v) of the 2-D image space ƒ(x, y) at v = 0:
P(u) = F(u,v)│v=0 (9)
Fig. 13. Geometrical aspects of the generation of transmission profiles. The Cartesian coor-
dinates (x, y) apply to the image space, ƒ. The coordinates that apply to the projection, p, are t,
being the distance from the projected X ray to the isocentre, and θ, being the projection angle.
This conclusion can be generalized for any projection angle θ and it thus provides the
proof for the central slice theorem. A reconstruction can thus, at least theoretically, be
achieved first by a construction of the 2-D Fourier space F(u,v) by many 1-D FT’s of the
projection profiles measured under many projection angles, and subsequently by a 2-D
inverse FT of the 2-D fourier space to the 2-D image space. The sampling of the 2-D
Fourier space from the 1-D FT’s of the projections yields a 2-D Fourier space in regular
polar coordinates. Prior to the 2-D inverse FT into image space, the regular distributed
points in the polar 2-D fourier space have to be transformed to regularly distributed points
in a cartesian 2-D Fourier space. The transformation from a polar coordinate system to a
cartesian coordinate system may lead to artifacts in the reconstructed image, owing to the
fact that the sampling of the 2-D Fourier space is denser around the origin (low
frequencies), and sparser further away from the origin (high frequencies) (Fig. 14).
20
Fig. 14. The CT scan yields a regular distributed sampling in polar coordinates of 2-D Fourier
space. Transformation into a regular distributed sampling in Cartesian coordinates is complicated,
particularly at higher frequencies (further from the origin).
A more accurate and practical reconstruction can be achieved with the formulation
known as the filtered back projection. the filtered back projection also starts with 1-D
FT’s of image space, thus creating the corresponding Fourier space, but the sampling
of the 2-D Fourier space F(u, v) is expressed on a polar grid using the coordinate
transform:
u = ωcosθ , v = ωsinθ (10)
The image reconstruction (the filtered back projection) is then expressed as:
f(x,y) = ∫ 𝑑𝜃𝜋
0∫ 𝑃(𝜔, 𝜃)│𝜔│𝑒𝑖2𝜋𝜔𝑡𝑑𝜔
+∞
−∞ (11)
where P(ω, θ) is the 1-D FT of the 1-D projection at angle θ, and |ω| is known as a ramp
filter in the frequency domain.
In practice, different filters can be used in the reconstruction, depending upon the
image properties required. The filter (or convolution kernel) in a filtered back projection
that theoretically yields an optimal reconstruction is the so-called Ramachandran–
Lakshminarayanan filter, also called the Ram–Lak or ramp filter. It provides optimal
spatial resolution in the reconstructed images. However, it also yields relatively high
noise levels in the reconstructed images. Such a theoretically ‘optimal’ filter in clinical
practice is referred to as a sharp or bone filter. Often, filters are used that reduce the
noise level in the reconstructed images; these filters provide some roll-off at higher
frequencies.
21
A modest roll-off is achieved with the Shepp–Logan filter, which provides images
that are less noisy and that provide better low contrast resolution and slightly inferior
spatial resolution in the reconstructed images; such filters are referred to as normal
filters. Even stronger roll-off at higher frequencies leads to further noise reduction,
better low contrast resolution, but noticeably poorer spatial resolution. Such filters in
clinical applications are referred to as soft tissue filters. CT scanners offer many
reconstruction filters that are optimized for specific clinical purposes. It is possible to
reconstruct one single CT scan with different reconstruction filters, in order to optimize
the visualization of, for example, both bone and soft tissue.
Other reconstruction techniques such as algebraic or iterative reconstruction can also
be used in CT. An algebraic reconstruction may seem attractive; however, algebraic
reconstruction through equation solving is not feasible in clinical practice, owing to the
large (512 × 512) matrices that are used in medical imaging and to inconsistencies in
the equations from measurement errors and noise.
Iterative (statistical) reconstructions are now commonly used in CT. The iterative
reconstruction is well known in medical imaging, since it is routinely used in nuclear
medicine. Iterative techniques provide potential benefits in CT, including the removal
of streak artefacts (particularly when fewer projection angles are used), and better
performance in low dose CT acquisitions. However, iteratively reconstructed images
may be affected by artefacts that are not present in filtered back projection images, such
as aliasing patterns and overshoots in the areas of sharp intensity transitions. Iterative
reconstruction algorithms are becoming popular in commercial CT scanners and can
produce low noise images.
1.11. Axial CT scan
An axial CT scan involves an acquisition of transmission profiles with a rotating X-
ray tube and a static table. An axial acquisition is generally performed with one full
360º rotation of the X ray tube, but to enhance temporal resolution this may be reduced
to a shorter ‘180º + fan angle’ acquisition. The rotation angle can be extended to, for
example, a 720º acquisition to enhance low contrast resolution by allowing a higher
tube charge (mAs). A complete CT scan generally involves subsequent axial
acquisitions in order to cover a clinically relevant volume. This is achieved by
translation of the table (‘step’) after each axial acquisition (‘shoot’).
22
Fig. 15. Scan projection radiographs for planning, respectively, CT brain, CT chest and CT
lumbar spine scans. The technician selects from the CT radiograph the optimal scan range, FOV
(marked in yellow) and angulation (head only).
Fig. 16. The SPR can be used to achieve AEC during the CT scan. The mAs values are indicated
at four levels, but during the helical acquisition, the tube charge is continuously optimized at
each level within the scanned range.
This is referred to as a step and shoot acquisition. Usually, the table translation is equal
to the slice thickness, so that subsequent axial acquisitions can be reconstructed as
contiguous axial images. Figure 17 (left) shows the geometry of an axial CT acquisition.
1.12. Helical CT scan
The helical CT scan was introduced in 1989, whereby the acquisition with a rotating X-
ray tube was combined with a moving table. The introduction of helical CT scans
improved the performance of CT considerably. Advantages of helical CT scans include
shorter scan time and more consistent 3-D image information for the scanned volume.
23
Disadvantages of helical CT scans include the introduction of artifacts such as the
windmill artifact. Figure 17 shows the geometry of a helical CT acquisition (right). The
circular trajectory of the X-ray tube transforms into a helical course from the perspective
of the patient. Helical scanning allows for the acquisition of a large volume of interest
within one breath hold and was a prerequisite for the development of high quality CT
Angiography. The table translation is generally expressed relative to the (nominal) beam
width (in single slice CT this equals the slice width): the ratio of table translation per 360°
tube rotation relative to the nominal beam width in helical CT is referred to as the pitch
factor. The rotation time of single slice CT scanners is 1–2 s and the slice thickness (and
nominal beam width) in most clinical applications is 5–10 mm.
Fig. 17. Geometry of an axial CT acquisition (left) and a helical CT acquisition (right).
1.13. MultiDetector CT (MDCT) scan
Ten years after the introduction of helical CT, the next step in CT technology that
provided even more new clinical applications was taken: the introduction of fast
rotating MDCT scanners with up to 64 adjacent active arrays of detectors, enabling the
simultaneous measurement of a correspondingly large number of transmission profiles.
At the same time, the rotation time dropped to 0.3–0.4 s, making it possible to scan
almost the entire body of an adult within one breath hold at a slice thickness well below
1 mm. Acquisitions with MDCT scanners are usually obtained in helical mode.
Exceptions include high resolution CT of the lungs, and step and shoot cardiac CT for
either coronary calcium scoring or coronary CT angiography.
24
1.14. Contrast Enhanced CT
Contrast can be artificially created within or between structures that would not be
visible on non-enhanced scans. For example, in CT Angiography, or CT Perfusion,
iodine is administered intravenously during the CT scan to enhance the contrast
between the vessel and the vessel walls (Fig. 18 (left)). In certain studies of the
abdomen, a diluted iodine solution is administered orally prior to the scan to enhance
contrast within the gastrointestinal tract. In CT colonography, gas may be inflated
through the rectum to enhance contrast between the colon and its surrounding tissues
(Fig. 18 (right)).
Fig. 18. CT angiography. 3-D rendered image of vessels with iodine administered intravenously
(left). CT colonography using gas inflated through the rectum (right).
1.15. Special Applications
Special applications of CT include the well established use for radiotherapy treatment
planning and for more experimental applications such as dual energy CT imaging and
dynamic volumetric CT studies.
Some scanners allow dynamic CT imaging (also known as 4-D CT), where a volume
of interest can be followed as a function of time. Such studies can be used to visualize
the movement of joints or the contrast enhancement of organs (Perfusion or dynamic
CT Angiography). Figure 19 shows an example of a dynamic CT Angiography study
of the entire brain with a volumetric CT scanner.
25
Fig. 19. A dynamic CT angiography study of the brain with a volumetric CT scanner that covers
the entire brain (Aquilion ONE, Toshiba).
In these images, time resolved contrast enhancement of the vessels allows the arterial
and venous phases to be followed. Other anatomical sites for CT perfusion studies
include the heart and the liver. During dynamic CT studies, as with CT fluoroscopy,
the operator should be aware that the skin dose might accumulate rapidly. The patient
skin dose should be maintained below 2 Gy to avoid deterministic skin effects such as
erythema and epilation.
1.16. CT image quality
The main acquisition parameters in CT are tube voltage, tube current and rotation
time. A relatively high tube voltage (120–140 kV) is used to achieve good X-ray
transmission and sufficient detector signal. For special applications, such as contrast
enhanced studies and paediatric examinations, it might be advantageous to use a
relatively low tube voltage, in the range 80–100 kV. The tube current used is limited
by the long scan time and the heat capacity of the X-ray tube, and by patient dose
considerations. To avoid motion artifacts, the rotation time needs to be as short as
possible. For scans that are less prone to motion artefacts and that require good low
contrast resolution (such as scans of the brain), a longer rotation time may be selected
to allow appropriate low contrast resolution. The excellent low contrast resolution of
CT images is the most prominent characteristic that distinguishes the CT modality from
other forms of non-tomographic radiography. Low contrast resolution is the ability to
detect structures that offer only a small difference in signal compared with their direct
26
environment. Image noise is the main limitation for low contrast resolution and may be
decreased with improved image quality by employing a number of strategies. Most
commonly, noise is reduced by increasing photon flux, which is achieved by increasing
the tube current (mA) at the cost of patient exposure. Alternatively, noise is reduced by
increasing the reconstructed slice thickness or by changing the selection of the
reconstruction algorithm, but at the cost of spatial resolution. Parameters that influence
the low contrast resolution include tube voltage, beam filtration and the use of a contrast
agent. The effect of noise in a CT image is seen in Fig. 20, where the 100% image
corresponds to an actual clinical acquisition. the raw data of the clinical acquisition
have been processed with a low dose simulation algorithm that adds noise to simulate
image quality for acquisitions that are performed at 75%, 50% and 25% of the clinically
used tube current. The appearance of the low contrast lesions in the liver becomes worse
at lower tube currents, owing to increased noise in the images.
Fig. 20. A contrast enhanced CT scan of the liver obtained at normal exposure (100%) and with
the addition of extra noise to simulate exposures of 75%, 50% and 25% of the normal exposure
Physicists usually test low contrast resolution performance using phantoms that
contain different sized low contrast inserts. Evaluation of the resultant image can be
either subjective, with an observer determining whether an insert is visible or not, or
objective, with calculation of the contrast to noise ratio. Determination of the noise
power spectrum would provide a more objective measure of scanner performance but
is not as yet applied on a large scale.
Spatial resolution, or high contrast resolution, is the ability to observe contours of
small objects within the scanned volume. Small objects can only be resolved well when
they exhibit a rather large difference in signal. Spatial resolution is limited by the
27
acquisition geometry of the CT scanner, the reconstruction algorithm and the
reconstructed slice thickness. Voxel size is often used as an indicator of spatial
resolution, although a smaller voxel size does not necessarily imply better spatial
resolution. Spatial resolution along the z-axis is usually determined using a slice
sensitivity profile, with the response often quantified as the full width at half maximum
(FWHM), while in the axial plane it is preferably measured as a point spread function
(PSF). From this, the modulation transfer function (MTF) can be calculated. The MTF
does yield useful information on image quality, although clinical assessment of the
MTF in the clinical environment may be complex and is usually only performed by
medical physicists at acceptance and commissioning of new scanners or after major
upgrades. Manufacturers of CT scanners provide information about the MTF, which
should be measured according to international standards. The performance of current
64 slice scanners with regard to spatial resolution, expressed as the full width at half
maximum of the PSF, is within the range 0.6–0.9 mm in all three dimensions.
Figure 21 shows images from a CatPhan phantom that is widely used to evaluate the
image quality of CT scans. The image on the left allows evaluation of the HU values
for four large inserts in the periphery of the phantom for the materials air, low density
polyethylene, PMMA and Teflon. Low contrast acrylic inserts of different diameters
around the centre are used for determining the effect of object size on low contrast
detectability. The image in the middle shows high contrast line pairs that allow the
subjective assessment of spatial resolution. The image on the right allows spatial
resolution to be measured objectively, as the PSF of a small tungsten bead. The image
can also be used to assess the homogeneity of the image.
Temporal resolution is the ability to resolve fast moving objects in the displayed CT
image. Good temporal resolution avoids motion artefacts and motion induced blurring
of the image. Good temporal resolution in CT is realized by fast data acquisition
through the fast rotation of the X ray tube. Reconstruction algorithms that are used for
general CT applications provide, in principle, a temporal resolution equal to the time of
a 360º rotation with full reconstruction.
The best routinely achievable temporal resolution is slightly longer than 50% of the
rotation time using 180º and fan angle rotation reconstruction. Temporal resolution can
be improved further by using dedicated reconstruction algorithms, for example, in
cardiac CT with a segmented reconstruction, or by using a dual source CT scanner.
28
There are no simple methodologies available yet that allow measurement of temporal
resolution in a clinical setting.
Fig. 21. Images of a CatPhan phantom taken at different z coordinates and showing different
modules of the phantom.
1.17. Effect of acquisition and reconstruction parameters on image quality
Many reconstruction and viewing parameters have an effect on image quality and
observer performance. These include the reconstructed slice thickness, the
reconstruction filter, the windowing and the image reformats that can be used in
addition to the review of axial images. Any CT acquisition can be reconstructed with
one or more reconstruction filters. Figure 22 shows the same image reconstructed with
slice thicknesses of 10, 5 and 0.5 mm. Note that in both the volume rendered and
coronal images, the spatial resolution improves considerably at smaller slice thickness.
Reconstructions are generally made at a slice thickness of ≤1 mm. During image
reading, the radiologist can choose the appropriate window settings for the specific
anatomy and pathology of interest. This is illustrated in Fig. 23 for four axial CT head
images all created by post-processing and derived from the same acquisition. Images
on the left are reconstructed with a soft tissue reconstruction filter; those on the right
are reconstructed with a sharp bone reconstruction filter. The images in the upper row
are shown in a window setting for brain (window level 50, window width 100); the
images in the lower row are shown in a window setting for bone (window level 1000,
window width 2500). As can be seen, the image at the top left is processed and
windowed appropriately for evaluation of the brain tissue. Likewise, the details in the
skull are better presented in the lower right image, owing to appropriate reconstruction
and settings. The image at the top right is hampered in its presentation of brain tissue,
owing to image noise that results from the inappropriate use of a bone reconstruction
29
filter, while the bone in the image cannot be assessed, owing to the window setting
used. Similarly, the image at the bottom left is hampered for bone analysis, owing to
blurring of the bone that results from the soft tissue reconstruction filter, while the brain
tissue cannot be assessed, owing to the use of the bone window setting. Many image
reformats can be used in addition to the reading of axial images. Figure 24 shows an
axial image of the brain and three additional reformats: a coronal image, a sagittal image
and a volume rendered image. Figure 25 shows two images of the chest, on the left a
maximum intensity projection, and on the right a 3-D volume rendered image.
Fig. 22. Reconstructions of the same acquisition at slice thicknesses of 10 mm (top), 5 mm
(middle) and 0.5 mm (bottom), on the left volume rendered images (3-D representation of the
study), on the right coronal views (2-D representation of the study).
Fig. 23. Four CT head images. Images on the left are reconstructed with a soft tissue
reconstruction filter; images on the right are reconstructed with a bone reconstruction filter.
30
Fig. 24. An axial image of the brain (top left) and three additional reformats: a coronal image
(top right), a sagital image (bottom left) and a volume rendered image (bottom right).
Fig. 25. Two images of the chest. On the left a maximum intensity projection and on the
right a 3-D volume rendered image.
1.18. Artifacts
Proper image quality in CT is only achieved if calibrations of the scanner are regularly
carried out according to the protocols that are prescribed by the manufacturers. These
include frequent air calibrations and, less frequently, calibrations with homogeneous
water phantoms. Air calibrations provide information about the small differences in the
response of individual dels. This is essential since the projections have to be accurate
to within 0.5%. Calibrations with phantoms allow some correction of the beam
hardening effect.
Artifacts can be acquisition related, reconstruction related, or patient related.
Acquisition related artefacts include ring artefacts (Fig. 26), which occur with one or
more malfunctioning dels, and unusable images, which may result from malfunctioning
31
of the X-ray tube during the acquisition. Undersampling the projection data may lead
to Moiré patterns, and detector afterglow may induce blurring of the image.
An important artefact that occurs when thick acquisition slices are used results from
an averaging of the linear attenuation coefficient along the z axis of a voxel. This artifact
is referred to as the partial volume effect; it will make a small high density object appear
as a larger lower density object and is seen, for example, when looking at cortical bone
in thick CT slices. This can be avoided by the use of thinner slices.
Strong attenuation of the X-ray beam by compact bone, calcifications or a metal
object may lead to a beam hardening artefact. A very severe artefact occurs when a
metal prosthesis is scanned; this effect is referred to as a metal artefact and occurs when
the prosthesis attenuates the X ray beam almost completely. Figure 26 also shows the
typical streaks in the axial image that occur when a large metal implant is scanned. In
this image, a hip prosthesis was scanned.
Fig. 26. A ring artifact occurs in the case of malfunctioning of one or more dels (left). Metal
artifacts occur as a result of beam hardening and low dose on the detector (right).
Patient related artifacts can sometimes be avoided by properly instructing the patient to
refrain from moving during the scan and to maintain the breath holding during the entire
scan, particularly during scans of the trunk. Movement of the heart and pulsation of the
vessels cannot be avoided. Therefore, it is essential that acquisitions of, for example,
the coronary arteries or the aorta are optimized to achieve the best possible temporal
resolution. It is well known that pulsation of the aorta may induce artefacts that mimic
an aortic dissection. In this case, the artifact may have serious consequences for the
patient if it is not recognized as an artifact (See Reference 35).
32
Chapter 2. CT dose descriptors
2.1. Radiation Dose Measures: General Definitions
Exposure
The term exposure describes the ability of x-rays to ionize air. It is measured in
roentgens (R); this unit is defined as the quantity of x rays that produces 2.580 ∙ 10-4 C
of charge collected per unit mass (kilograms) of air at standard temperature and pressure
(STP): 1 R = 0.000258 C/kg air. This term refers to the concentration, in air, of radiation
at a specific point and is the ionization produced in a specific volume of air. It is
typically measured with an ionization chamber and an electrometer. It essentially
describes how much ionization is present in the volume, but it does not tell how much
energy is absorbed by the tissues being irradiated.
X =dQ
dm (12)
Absorbed Radiation Dose
Absorbed radiation dose, often referred to as radiation dose, describes the amount of
energy absorbed per unit mass at a specific point. It is measured in grays (1 Gy = 1
J/kg) or rads (1 rad = 100 erg/g). The conversion between rads and grays is 100 rad = 1
Gy. Absorbed dose essentially describes how much energy from ionizing radiation has
been absorbed in a small volume centered at a point; it does not describe where that
radiation dose is absorbed or reflect the relative radiosensitivity or risk of detriment to
those tissues being irradiated.
D =dE
dm (13)
Effective Dose
CTDI and DLP are CT-specific dose descriptors that do not allow for comparisons
with radiation exposures from other sources, e.g. projection radiography, nuclear
medicine or natural background radiation. The only common denominator to achieve
this goal is the ‘Effective Dose’. Effective dose takes into account where the radiation
dose is being absorbed (e.g. which tissue has absorbed that radiation dose) and attempts
33
to reflect the equivalent whole-body dose that results in a stochastic risk that is
equivalent to the stochastic risk from the actual absorbed dose to those tissues irradiated
in a nonuniform, partial-body irradiation such as a CT scan. It is a weighted average of
organ doses, as described in Equation (14):
ED = ∑T (wT ∙ wR ∙ DT,R) (14)
where ED is the effective dose, wT is the tissue weighting factor (see Table 3), wR is
the radiation-weighting coefficient (1 for x rays) (see Table 4), DT,R is the average
absorbed dose to tissue T, T is the subscript for each radiosensitive tissue, and R is the
subscript for each type of radiation. The weighting factors are set for each radiosensitive
organ in Publication 103 of the International Commission on Radiological Protection
(ICRP 103-2007).
Table 3. Table 4.
Effective dose cannot as such be measured directly in vivo. Measurements in
anthropomorphic phantoms with thermoluminescent dosImeters (TLD) are very time-
consuming and therefore not well suited for daily practice. Effective dose, however,
can be assessed in various ways by using conversion factors. For coarse estimates, it is
sufficient to multiply the dose-length product with mean conversion factors, depending
on which one out of three body regions was scanned and whether the scan was made in
head or body scanning mode with the following equation :
ED = k ∙ DLP (15)
Effective dose is measured in sieverts (Sv) or rems. The conversion between sieverts
and rems is 100 rem = 1 Sv. Although methods to calculate the effective dose have been
established, these methods depend heavily on the ability to estimate the dose to
34
radiosensitive organs from the CT procedure (DT,R). Care is also needed to not mix up
effective dose with organ doses, as both are expressed in mSv. Nevertheless, effective
dose is of great value, e.g. to answer questions raised from patients. For this purpose,
the annual natural background radiation, which is between 2 and 3 mSv in many
countries, can be used as a scale.
2.2. Radiation Dose Measures: CT Specific
Because of its geometry and usage, CT is a unique modality and therefore has its own
set of specific parameters for radiation dose. This modality is unique because the
exposure is essentially continuous around the patient, rather than a projectional
modality in which the exposure is taken from one or two source locations. The modality
typically uses thin sections (ranging from 0.5-mm to 20-mm) nominal beam
collimation. However, this modality also typically uses multiple exposures along some
length of the patient to cover a volume of anatomy. In addition, these exposures may be
done in sequences of scans (e.g. a series of scans such as pre- and postcontrast).
Variations within the Scan Plane
Projectional radiographic exposures are taken from one source position and the
entrance skin dose is much larger than the exit skin dose, creating a large radiation dose
gradient across the patient (Fig. 27) as the dose decreases continuously from the
entrance of the X-ray beam to its exit, with a ratio of between 100 and 1000 to 1. In
contrast, the tomographic exposure of CT scans with a full 360° rotation equally
irradiates the patient from all directions resulting in a radially symmetric radiation dose
gradient within the patient. Therefore, a dose comparison of CT with conventional
projection radiography in terms of skin dose doesn’t make any sense. Thus, in a uniform
circular object, such as a test phantom, all of the points at a certain radius from the
center have the same (or nearly the same) radiation dose (Fig. 28). As we shall see, the
magnitude of that dose gradient (the size of the difference from center to periphery)
will be affected by several factors, including the size of the object, the x-ray beam
spectrum, and the attenuation of the material or tissue. For example, in a typical CT
dosimetry phantom that is 32 cm in diameter and made of polymethyl methacrylate
(PMMA) (usually referred to as the body phantom) measurements of CT dose, which
35
will be defined later, obtained at the center are typically about 50% of the measured
value obtained at one of the peripheral positions.
Fig. 27. Dose gradient resulting from Fig. 28. Dose gradient resulting from a full
a projectional radiographic exposure 360° exposure from a CT scan. The thicker
in which the source is stationary at lines represent the entrance skin dose, which
one position. The thicker lines repre- is much larger than the dose at the inner
sent the entrance skin dose, which is radius, represented by the thinner lines. This
much larger than the exit skin dose, difference results in a radially symmetric
represented by the thinner lines. radiation dose gradient within the patient.
This difference creates a linear
gradient through the patient.
This result is illustrated in Figure 29, which shows the center value obtained under
specific conditions to be approximately 10 mGy while the peripheral values are 20 mGy
under those same conditions. However, for a smaller-diameter phantom (the 16-cm-
diameter phantom referred to as the head phantom) measured under the identical
exposure conditions, the center value reading climbs to approximately 40 mGy, as do
the peripheral values (Fig 30). This indicates that the magnitude of the difference from
center to periphery is very much size dependent; it also indicates that the absolute
values of the absorbed doses are size dependent. For the phantoms see Figure 31.
36
Fig. 29. Typical dose measurements in a Fig. 30. Typical dose measurements in a 16-
32-cm-diameter (body) phantom from a cm-diameter (head) phantom from a CT
single-detector CT scan. Values measured scan. Values measured at the center and pe-
at the center and periphery (1 cm below the riphery (1 cm below the surface) positions
surface) positions within a polymethyl within a polymethyl methacrylate circular
methacrylate circular dosimetry phantom dosimetry phantom demonstrate essentially
demonstrate a radial dose gradient with a 2:1 no radial dose gradient. Technical factors
ratio from periphery to center. Technical fac- for the measurements were 120 kVp, 300
tors for the measurements were 120 kVp, 280 mA, 1-sec scan (ie, 300 mAs), and 5-mm
mA, 1-sec scan (ie, 280 mAs), and 10-mm collimation.
collimation.
Figure 31.
Z-Axis Variations
In addition to the variations within the scan plane, there are variations along the length
of the patient or phantom. These can be characterized by the z-axis dose distribution or
radiation profile (Fig. 32). This is the distribution of absorbed dose along the axis of
the patient due to a single axial scan (a full rotation at one table position).
Fig. 32. Radiation profile of a full-rotation CT scan measured at isocenter. This profile is the
distribution of radiation dose along the axis of the patient (the z axis) and is known as D(z).
37
The radiation profile is not limited to the primary area being imaged, and there are
tails to this distribution from the non-ideal collimation of the x-ray source and from
scatter of photons within the object being exposed. When multiple adjacent scans are
performed, the scanning procedure using narrow beams along the longitudinal z-axis of
the patient implies that a significant portion of the radiation energy is deposited outside
the nominal beam width and the tails of the radiation profiles from adjacent scans can
contribute to the absorbed dose outside of the primary area being imaged. This is mainly
due to penumbra effects and scattered radiation produced inside the beam. If these tails
are significant and are non-zero at some distance from the location of the originating
section, then these contributions can add up, creating additional absorbed dose in the
primary area being imaged. That is, the radiation dose in a specific section consists of
the sum of contributions to that section when that area is the primary area being imaged
as well as the contributions from the tails of radiation profiles from adjacent sections
when other locations are the primary area being imaged. Thus, the average level of the
total dose profile, which is called ‘Multiple Scan Average Dose (MSAD)’, is higher
than the peak value of each single dose profile. This increase results from the tails of
the single dose profiles for a scan series. (Fig. 33).
Figure 33.
38
The size of the contributions from adjacent sections is very directly related to the
spacing of sections and the width and shape of the radiation profile. To account for the
effects from multiple scans, several dose descriptors were developed. One of them was
the Multiple Scan Average Dose (MSAD) descriptor. This is defined as the average
dose resulting from a series of scans over an interval I in length:
(16)
where I is the interval of the scan length and Dseries(z) is the dose at position z parallel
to the z (rotational) axis resulting from the series of CT scans. Following this was the
Computed Tomography Dose Index (CTDI). This was defined as the radiation dose,
normalized to beam width, measured from 14 contiguous sections:
(17)
where n is the number of sections per scan, T or hcol is the width of the interval equal to
the selected section thickness, and Dsingle(z) is the dose at point z on any line parallel to
the z (rotational) axis for a single axial scan. Obviously, MSAD and CTDI are exactly
equal if the table feed TF or I is equal to the nominal beam width N·T or N·hcol, i.e. if
the pitch factor :
pitch = I / N·T = TF / N·hcol (18)
is equal to 1. CTDI is therefore equal to the area of the dose profile (the ‘dose profile
integral’) divided by the nominal beam width. (Fig. 34).
Fig. 34. ’Computed Tomography Dose Index (CTDI)’: CTDI is the equivalent of the dose value
inside the irradiated slice (beam) that would result if the absorbed radiation dose profile were
entirely concentrated to a rectangular profile of width equal to the nominal beam width N·hcol.
39
However, to be measured according to the definition, only 14 sections could be
measured and one had to measure the radiation dose profile (typically done with
thermoluminescent dosimeters (TLDs) or film, neither of which was very convenient).
Measurements of exposure could be obtained with a pencil ionization chamber, but its
fixed length of 100 mm meant that only 14 sections of 7-mm thickness could be
measured with that chamber alone. To measure CTDI for thinner nominal sections,
sometimes lead sleeves were used to cover the part of the chamber that exceeded 14
section widths. To overcome the limitations of CTDI with 14 sections, another radiation
dose index (CTDI100) was developed. This index relaxed the constraint on 14 sections
and allowed calculation of the index for 100 mm along the length of an entire pencil
ionization chamber (Fig. 35), regardless of the nominal section width being used.
Figure 35.
These detectors accumulate the dose profile integral (DPI), i.e. the area under the dose
profile shown in Fig. 6. The CTDI is then obtained according to equation 3 by division
with the nominal beam width. This index is therefore defined as follows:
(19)
where N is the number of acquired sections per scan (also referred to as the number of
data channels used during acquisition) and T is the nominal width of each acquired
section (which is not necessarily the same as the nominal width of the reconstructed
section width). Thus, the exposure measurement, performed with one axial scan either
in air or in one of the polymethyl methacrylate standard dosimetry phantoms with
patient-like diameters for which CTDI is defined, results in a calculated dose index,
CTDI100. This index can be measured and calculated for the center location as well as
at least one of the peripheral positions (1 cm below the surface) within the phantom to
40
describe the variations within the scan plane as well. The larger phantom, being 32 cm
in diameter, represents the absorption that is typical for the trunk region of adults. The
smaller phantom (16 cm in diameter) represents the patient in head examinations. The
smaller phantom is also used for dose assessment in paediatric examinations.
CTDIw was created to represent a dose index that provides a weighted average of the
center and peripheral contributions to dose within the scan plane and represents the
CTDI averaged over the cross section of the pertaining phantom. This index is used to
overcome the limitations of CTDI100 and its dependency on position within the scan
plane. The definition is as follows:
(20)
One final CTDI descriptor takes into account the parameters that are related to a specific
imaging protocol, the helical pitch or axial scan spacing, and is defined as CTDIvol:
CTDIvol = CTDIw ∙ NT/I (21)
where N and T are as defined earlier and represent the total collimated width of the x-
ray beam and I is the table travel per rotation for a helical scan or the spacing between
acquisitions for axial scans. For helical scans, the following formula is being used:
CTDIvol = CTDIw / pitch (22)
where pitch is defined as table distance traveled in one 360° rotation/total collimated
width of the x-ray beam. Since averaging includes both the cross section and the scan
length, CTDIvol therefore represents the average dose for a given scan volume. CTDIvol
is used as the dose quantity that is displayed at the operator’s console of newer scanners.
Another dose descriptor that is related to CTDI and is commonly reported on CT
scanners and in the literature is the dose-length product (DLP). DLP takes both the
‘intensity’ (represented by the CTDIvol) and the extension (represented by the scan
length L) of an irradiation into account (Fig. 36). This value is simply the CTDIvol
multiplied by the length of the scan (in centimeters) and is given in units of
milligray∙centimeters:
DLP = CTDIvol ∙ Scan Length (23)
41
Fig. 36. Total dose profile of a scan series with subsequent rotations. The dose-length product
(DLP) is the product of the height (dose, i.e. CTDIvol) and the width (scan length L) of the total
dose profile and is equal to the area under the curve.
Thus, the dose-length product increases with the number of slices (correctly: with the
length of the irradiated body section), while the dose (i.e. CTDIvol) remains the same
regardless of the number of slices or length, respectively. In Fig. 36, the area of the total
dose profile of the scan series represents the DLP. DLP is the equivalent of the dose-
area product (DAP) in projection radiography, a quantity that also combines both
aspects (intensity and extension) of patient exposure.
In sequential scanning, the scan length is determined by the beam width N∙T or N·hcol
and I or the number n of tables feeds TF:
L = n ∙ TF + N·hcol (24)
while in spiral scanning the scan length only depends on the number n of rotations and
the table feed TF:
L = n ∙ TF = N ∙ hcol ∙ p ∙ T / trot (25)
where T is the total scan time, trot is the gantry rotation time, and p is the pitch factor.
While in sequential scanning the scan length L is equal to the range from the begin of
the first slice till the end of the last, the (gross) scan length for spiral scanning not only
comprises the (net) length of the imaged body section but also includes the additional
rotations at the begin and the end of the scan (‘overranging’) that are required for data
interpolation.
If an examination consists of several sequential scan series or spiral scans, the dose-
length product of the complete examination (DLPexam) is the sum of the dose-length
products of each single series or spiral scan:
42
DLPexam = ∑ DLP (26)
This descriptor is used in one approach to obtain an estimate of effective dose that
will be described later.
These CTDI descriptors are obviously meant to serve as an index of radiation dose
due to CT scanning and are not meant to serve as an accurate estimate of the radiation
dose incurred by an individual patient. Although the phantom measurements are meant
to be reflective of an attenuation environment somewhat similar to a patient, the
homogeneous polymethyl methacrylate phantom does not simulate the different tissue
types and heterogeneities of a real patient.
2.3. Factors That Influence Radiation Dose from CT
In general, there are some factors that have a direct influence on radiation dose, such
as the x-ray beam energy (kilovolt peak), tube current (in milliamperes), rotation or
exposure time, section thickness, object thickness or attenuation, pitch and/or spacing,
dose reduction techniques such as tube current variation or modulation, and distance
from the x-ray tube to isocenter. In addition, there are some factors that have an indirect
effect on radiation dose (those factors that have a direct influence on image quality, but
no direct effect on radiation dose); for example, the reconstruction filter. Choices of
these parameters may influence an operator to change settings that do directly influence
radiation dose. These factors are discussed in this section.
Beam Energy
The energy of the x-ray beam has a direct influence on patient radiation dose. This is
selected by the operator (technologist) when the kilovolt peak is chosen for the scan.
However, it is also influenced by the filtration selected for the scan. On some scanners,
the selection of filtration is explicit; for others, it is implied (e.g. by selection of the
scan field of view [SFOV]). The influence of beam energy is shown in Table 5. When
all other technical parameters are held constant and the kilovolt peak is increased on a
single-detector CT scanner, the CTDIw values also increase for both the head and body
CTDI phantoms. For example, when the kilovolt peak was increased from 120 to 140
on a CT, the CTDIw increase was 37.5% for the head phantom and 39% for the body
phantom.
43
Photon Fluence
The photon fluence, as influenced by the tube current–time product (milliampere-
seconds), also has a direct influence on patient radiation dose. As one might expect, the
radiation dose is directly proportional to the milliampere-seconds value. This is shown
in Table 6, which gives the results when the milliampere-seconds value is increased
and all other technical parameters are held constant on a single-detector CT scanner.
Under these conditions, the CTDIw values increase linearly with milliampere-seconds
for both the head and body CTDI phantoms.
Note that these results hold only while the tube current–time product is varied and all
other parameters are held constant. This is an issue because on some scanners, the user
inputs a parameter labeled “mAs,” but that parameter is really the effective milliampere-
seconds value, which is milliamperage ∙ time/pitch. On these scanners, when pitch is
varied, the milliampere-seconds value is varied in a corresponding fashion to keep the
effective milliampere-seconds value constant.
44
Helical Pitch
For helical scans, the pitch parameter (defined as table distance traveled in one 360°
rotation/total collimated width of the x-ray beam) has a direct influence on patient
radiation dose. This is essentially because as pitch increases, the time that any one point
in space spends in the x-ray beam is decreased. The relationship between radiation dose
and pitch has been shown previously by using phantoms and thermoluminescent
dosimeters. On the basis of these results, the CTDIvol (which is the only CTDI descriptor
that takes pitch into account) varies as shown in Table 7, which gives the results when
the pitch is varied and all other technical parameters are held constant on a single-
detector CT scanner.
X-ray Beam Collimation: Single-Detector Scanners
The collimation of the x-ray beam will both directly and indirectly influence the patient
radiation dose. The indirect effects will be described later. For a single section with all
other technical parameters held constant, more x-ray photons will be transmitted when
the collimator setting is wider (wider x-ray beam for a thicker section). However,
exposure and absorbed radiation dose are defined on a per unit mass basis. The thicker
section has more photons available but also more mass being irradiated than a thinner
section, thus indicating that the radiation dose for thick and thin sections may be close
to equivalent (the difference might be attributed to the higher scatter expected in the
thicker section). This equivalence would also assume that the radiation profiles (and, as
shown earlier, the overlap between adjacent exposures) are equivalent between narrow
45
and wide collimation settings. However, previous publications have shown that this is
not quite true for single-detector scanners and that thinner collimations typically result
in a greater degree of overlap and higher CTDI values. The results from measuring
CTDIw are shown in Table 8, which gives the results when the collimation is varied and
all other technical parameters are held constant on a single-detector CT scanner.
X-ray Beam Collimation: Multiple-Detector Scanners
Although the effects of beam collimation were small for a single-detector scanner,
current experience shows that this is not the case with multidetector scanners. In fact,
early reports from early versions of multidetector scanners showed significant
dependence on x-ray beam collimation. These effects result from differences in x-ray
beam collimation (even when the same reconstructed section thickness is used). That
is, on many multidetector scanners, there are several ways to scan and reconstruct
images that have the same section thickness. For example, on a multidetector CT
scanner, one can perform axial scans of 4 x 1.25 mm (5-mm beam width), 4 x 2.5 mm
(10-mm beam width), and 4 x 5 mm (20-mm beam width) to make a 5-mm-thick
reconstructed section. For each of these modes, when CTDIw values are measured, there
is a surprising difference in absorbed dose. These results are shown in Table 9, which
gives the results when all other technical parameters are held constant on that
multidetector CT scanner. These results show that the difference in beam collimation,
not the reconstructed section width, makes a significant difference in CTDIw. These
differences may be as much as 55% in the head phantom and 65% in the body phantom,
with the higher doses coming when narrower beam collimation is used.
46
Figure 37.
Figure 38.
47
Figure 39.
2.4. Effects of Object (and Patient) Size
In each of the preceding sections, we reported results for both the head and body
phantoms. These phantoms, as described earlier, are made of the same soft-tissue–
equivalent material but are 16-cm-diameter and 32-cm-diameter right circular
cylinders, respectively. To produce each table, we used the same technical factors for
each phantom. Therefore, the primary difference in results between the head and body
phantoms is size. Each of the tables shows that when the same technical parameters are
used, the appropriate index shows that the smaller object always absorbs the higher
dose and that the difference is at least a factor of two. Thus, for the same exposure
factors, smaller patients would be expected to absorb much higher amounts of radiation
dose than larger patients. This has significant implications for pediatric patients and
small adults.
This is primarily because tissues are being exposed with both entrance radiation (as
the tube is positioned directly over the tissue) and exit radiation (as the tube moves to
the other side of the patient) as the source moves around the patient. For smaller
patients, the exit radiation has been attenuated by less tissue and therefore is closer to
the entrance radiation in its intensity, resulting in a much more uniform dose
distribution (nearly equal at all locations in a 16-cm-diameter phantom). For the larger
patient, the exit radiation is much less intense due to its attenuation through more tissue.
This results in a difference within the scan plane with the higher radiation dose values
occurring near the periphery, where entrance exposure is highest.
The effect of patient or object size on radiation dose has brought significant
discussion into the proper selection of protocols for imaging pediatric patients as well
as adjusting technical factors for patients according to size.
48
2.5. Other Options for Reducing Scan Dose
In addition to the technical parameters discussed earlier, manufacturers have recently
provided users with other means to reduce patient dose. One of these is an option to
make changes in tube current based on the estimated attenuation of the patient at a
specific location. Thus, the tube current will be programmed to a maximum value and
can be reduced when there is information that a location along the patient is expected
to be less attenuating than the most attenuating location to be imaged. This is
determined by using both anterioposterior and lateral planning projection views. From
these views, the tube current will be programmed to vary by location along the length
of the patient and even as the tube is rotating around the patient. The exact details of
the option vary by manufacturer. In the near future, manufacturers may provide real-
time (or close to real-time) tube current modulation, so that tube current can be varied
(reduced) as the scan is actually occurring, eliminating the need for both planning
projections for dose reduction purposes (they may still be needed for planning
purposes).
2.6. Indirect Effects
In addition to the direct effects that collimation has, as described earlier, there are
some indirect effects that both it and the reconstruction algorithm may have on radiation
dose. This is because, when thinner reconstructed image thicknesses are used, with all
other factors held constant, there will be more noise in the image (where noise is defined
as the standard deviation of the CT number). Therefore, noise typically increases with
1/√T, where T is the nominal section thickness. Therefore, a 10-mm-thick section is
expected to have 3.2 times less noise than a 1-mm-thick section. Often when noisy
images are obtained, the kilovolt peak or milliampere-seconds value or both are
increased to offset the increase in noise due to narrower sections. Similar behavior is
observed for the effects of the reconstruction algorithm. Algorithms that enhance higher
spatial frequencies and improve spatial resolution (such as required for lung or skeletal
imaging) also increase the noise in the image. To overcome this increase in noise, the
kilovolt peak or milliampere-seconds value or both may be increased. This increase in
kilovolt peak or milliampere-seconds value will result in an increase in radiation dose.
Therefore, although changing the algorithm or section thickness may not have a direct
49
effect on radiation dose, the selection of technical factors to offset the resulting increase
in image noise may result in an increase in radiation dose.
2.7. Methods to Reduce Patient Dose
From the preceding discussion, it appears that there are several mechanisms to reduce
the radiation dose to a patient. However, each of them has some resulting trade-off
involved. Each of these is discussed below.
Reducing the Milliampere-Seconds Value
From the results presented earlier, the radiation dose is linear with the milliampere-
seconds value when all other factors are held constant. So, if the milliampere-seconds
value is reduced by 50%, the radiation dose will be reduced by the same amount.
However, this reduction will increase image noise by 1/√(mAs), which means that a
50% reduction in the milliampere-seconds value results in a noise increase of 41%
(1/√2=1.41), a 41% increase). Depending on the requirements of the clinical
application, this reduction may be readily accepted; in other cases, this type of reduction
in milliampere-seconds may compromise the diagnostic quality of the imaging
examination. For example, detection of high-contrast objects in the lung may not
require a low-noise imaging protocol and the reduction in milliampere-seconds may be
well tolerated. On the other hand, imaging low-contrast lesions in the liver does require
a low-noise imaging protocol and the reduction in milliampere-seconds may limit the
ability to detect these lesions.
Increasing Pitch
The radiation dose is inversely proportional to pitch when all other factors are held
constant. Therefore, increasing pitch is one consistent way to reduce radiation dose.
The trade-off in increasing pitch is an increase in effective section thickness, which
results in increased volume averaging, which in turn may reduce the image signal
(contrast between some object and background). The ability to use this type of dose
reduction again depends on the clinical application.
50
Varying the Milliampere-Seconds Value by Patient Size
CT is an example of a digital modality in which the image quality continues to
improve as the exposure increases. This is contrasted with analog projectional film, in
which too high of an exposure results in an overexposed (too dark) film. Thus, when
pediatric patients or small adult patients are imaged with CT using full-sized adult
techniques, there is no penalty to image quality; in fact, the image quality is better under
these conditions, as more photons reach the detector and image noise is reduced.
However, the radiation dose to the smaller patient is potentially higher than is necessary
to obtain a diagnostic image. Therefore, significant effort has recently been put into
developing size- and weight-based imaging protocols to reduce radiation dose to
pediatric patients and small adult patients, so that radiation dose can be reduced while
still achieving sufficient diagnostic image quality. This has typically been in the form
of a reduced milliampere-seconds value for reduced patient size and has led to the
development of suggested technique charts for pediatric patients.
Reducing Beam Energy
As discussed earlier, reducing the beam energy results in reduced radiation dose when
all other factors are held constant. This will increase the image noise, and contrast
changes will occur with a change in kilovolt peak, increasing with lower kilovolt peak
for most tissue interfaces but decreasing or changing very little for others. From
CTDI100 results, and not from the CTDIw results, we can observe that the radiation dose
gradient is larger from periphery to center in the body phantom at lower kilovolt peak
settings. This implies a greater relative skin dose for patients when lower kilovolt peak
settings are used (See References 33 and 34).
.
51
SPECIFIC PART
Chapter 3. CT imaging modalities
3.1. Theoretic basis
Stroke is a syndrome caused by disruption of the blood flow to part of the brain due
to either :
(a) occlusion of a blood vessel (ischemic stroke) or
(b) rupture of a blood vessel (hemorrhagic stroke)
resulting in injury to cells and causing sudden loss of focal brain functions, such as :
1. Confusion, trouble speaking or understanding speech
2. Numbness or weakness of face
3. Trouble seeing
4. Trouble walking, dizziness, loss of balance or coordination
5. Severe headache with no known cause
Ischemic stroke can be either thrombotic, embolic or lacunar. In the thrombotic case,
a clot has been formed in a vessel inside the human brain, causing the obstruction of
blood flow at the exact same location. Embolic is the case where a clot has been formed
elsewhere and through blood circulation has been transferred to the brain where it set
in and caused the disruption of blood flow in the region. In the lacunar case, there is an
occlusion of small cerebral arterioles.
Moreover, hemorrhagic stroke can be either intracerebral or subarachnoid. In the
first case of intracerebral (or intracranial - ICH) hemorrhage, there is bleeding present
in the brain parenchyma, whereas in the latter case of subarachnoid hemorrhage (SAH),
the bleeding is limited in the subarachnoid space and the cerebrospinal fluid.
In view of the updated guidelines for the endovascular treatment of ischemic stroke,
the only therapy for acute stroke currently approved by the U.S. Food and Drug
Administration and the European Union is intravenous thrombolysis. However, thrombolysis has the most impact in the first 6 hours from symptom onset, with the
benefit of intravenous thrombolysis decreasing steadily over time, resulting in an
interventional time window as narrow as 3 hours.
52
Thus, patients must be selected :
1. Accurately :
i) Since patients may present with similar to brain ischemia clinical
findings and
ii) In order to be able to exclude hemorrhage or other mimicking lesions
2. Timely :
Timely diagnosis is achieved by a comprehensive CT imaging protocol.
The comprehensive prescription protocol (CPP) about brain perfusion performed at the
University Hospital of Patras consists of three types of examinations; an unenhanced
brain CT (UN-CT), a brain CT Angiography (CTA) and a brain CT Perfusion (CTP)
scan.
3.2. Unenhanced brain CT (UN-CT)
The aims of performing an unenhanced brain CT (UN-CT) are :
3.2.1. Detection of hemorrhage or stroke mimickers (such as a neoplasm or an
arteriovenous malformation) that could be the cause of the neurologic deficit
Figure 40.
53
3.2.2. Distinction between Ischemia and Hemorrhage (Necessary but not sufficient to
complete the initial workup)
Figure 41.
3.2.3. Detection of ischemic signs of established infarction such as a cortical-
subcortical hypoattenuating area within a vascular territory
Figure 42. Drawings (top) illustrate the territories (blue) of the Anterior cerebral
artery (ACA), middle cerebral artery (MCA) and posterior cerebral artery (PCA).
CT scans (bottom) show established infarctions of these arteries.
54
i) The presence of hypoattenuation affecting more than one-third of
the MCA territory is a contraindication for revascularization
ii) It has been demonstrated that hemorrhagic complications are
associated with larger established infarcted lesions before
treatment
3.2.4. Time should not be lost in puzzling over subtle early ischemic signs since
the UN-CT holds a 15-60% sensitivity to acute ischemic changes in the
first 6h.
Figure 43.
3.2.5. Detection of early acute ischemic stroke on nonenhanced CT images may
be improved by using variable window width and center level settings to
accentuate the contrast between normal and edematous tissue
Guideline : 1st (W : 40 HU, C : 20 HU) / 2nd (W : 20 HU, C : 32 HU)
Figure 44.
55
3.2.6. Sensitivity (61%) was observed for depiction of subtle early signs of infarction
and ischemia, including :
(a) subtle hypoattenuation (lower arrow)
(b) cortical sulcal effacement (upper arrow)
Figure 45.
(c) hyperattenuation of a large vessel
(hyperattenuating MCA sign or dot sign)
Figure 46.
(d) obscuration and loss of gray matter-white
matter differentiation in the basal ganglia
Figure 47.
(e) loss of the insular ribbon
Figure 48.
56
3.3. Brain CT Angiography (CTA)
The main role of performing a brain CT Angiography (CTA) scan, is to reveal the status
of large cervical and intracranial arteries and thereby :
i) help define the occlusion site
ii) depict arterial dissection
iii) grade collateral blood flow
iv) characterize atherosclerotic disease
Brain CT Angiography is a Contrast Enhanced CT scan that :
3.3.1. Delineates the presence and site of vascular occlusion
Figure 49.
3.3.2. Depicts of arterial dissection
a. Intraarterial thrombolysis has been associated with higher recanalization rates
for occlusions of the internal carotid artery (ICA), MCA stem, basilar artery
Figure 50.
57
b. Brain CT Angiography is useful in detecting occlusions of the
internal carotid artery, MCA stem, basilar artery and differentiating
them from more distal (M2 or M3) occlusions for intravenous,
intraarterial, or mixed (intravenous-intraarterial) treatment planning
Figure 51.
c. Especially important for the detection of vertebrobasilar thrombo-
sis, since this entity is very difficult to detect at unenhanced CT
and the brainstem is frequently not included in perfusion coverage
Figure 52.
58
d. The main pitfalls are caused by basilar artery occlusions that are
missed because unenhanced CT and perfusion CT are performed, but
not CTA. It also detects the presence of a filling defect in the vessel
caused by arterial thrombosis with a sensitivity of 89% compared with
conventional angiography
Figure 53. Acute stroke. Findings at initial unenhanced CT were normal, and CT angiography was not
performed. (a) Follow-up unenhanced CT scan (36 hours evolution) shows a hypoattenuating midbrain
(arrowheads) and a hyperattenuat-ing basilar artery (arrow). (b) CT angiogram helps confirm a filling
defect of the basilar artery (arrowheads) related to pons infarction and basilar artery obstruction.
e. It is a thin-section examination performed with a time optimized
bolus of nonionic contrast medium (iodine: 300–400 mg/ml) to enhan-
ce the carotid and vertebral arteries and the circle of Willis. Evaluation
of the main intracranial arteries is performed with multiplanar
maximum-intensity projection (MIP) reformatting (thickness ~ 20
mm) on a three-dimensional workstation for rapid identification of
occlusion or stenosis of the carotid artery or MCA. Basic principle of
MIP: Parallel rays are cast through a VOI. The maximum CT number
along each projecting ray is displayed as an MIP image.
Figure 54.
59
Figure 55. Acute stroke (1.5 hours evolution) in a 57-year-old woman with right hemiplegia. (a)
Nonenhanced CT scan shows loss of the insular ribbon in the left MCA and hypoattenuation of the left
lenticular nucleus. (b–d) Perfusion CT maps of MTT (b) and CBV (c) and a summary map (d) show
extensive infarction with reduced mismatch. (e) Axial MIP reformatted CT angiographic image shows
left MCA obstruction (arrows). (f) CT angiographic–source image shows an area of hypoattenuation,
thereby helping confirm core infarction (orange dots).
3.3.3. Grades collateral blood flow
Figure 56.
Grading of collateral vessels in the territory of the occluded artery may be performed
by the 3 most commonly used scores :
I. a 4-point absolute score assessing the percentage of the ischemic bed in which
collateral filling is seen
II. a 5-point relative score comparing collateral vessel enhancement within the
ischemic bed with similar contralateral vessels
III. a 20-point detailed relative score comparing collateral vessel enhancement with
similar contralateral vessels
60
The 3 most commonly used scores are in detail:
I. 4-point absolute score
a) 0 = no visible collateral vessels to the ischemic site
b) 1 = visible collateral vessels to the periphery of the ischemic site
c) 2 = complete irrigation of the ischemic bed by collateral flow
d) 3 = normal antegrade flow
II. 5-point relative score
Table 10.
III. 20-point detailed relative score
i) ASPECTS methodology employs a 20-point grading scale
ii) It scores the collaterals in the 6 ASPECTS cortical regions (M1-6),
caudate, insular ribbon, internal capsule & lentiform nucleus
(lenticulostriate arteries) in the basal ganglia arising from retrograde
MCA filling
iii) It scores the contrast’s opacification extent in arteries distal to the
occlusion to form a score from 0-20
a) 0 = artery not seen
b) 1 = less prominent
c) 2 = equal or more prominent compared to a matching region in the
opposite hemisphere
C = caudate head, L = lentiform nucleus, IC = internal capsule, I = insular
ribbon, M1 = anterior MCA cortex, M2 = MCA cortex lateral to insular ribbon,
M3 = posterior MCA cortex, M4, M5, and M6 are anterior, lateral, and posterior
MCA territories, respectively, approximately 2 cm superior to M1, M2, and M3,
respectively, rostral to basal ganglia.
61
Figure 57.
Table 11.
All images are reconstructed with a section thickness of 20 mm to increase the
visualization of vessel continuity. Assessment of collateralization in conventional CT
angiography and temporal MIPs of the arterial, arteriovenous, and venous phases are
presented in the image below.
Figure 58.
62
For relative collateral grading in dynamic CT angiography, contralateral arterial
vessel status as a reference to prevent venous superimposition was used. Collaterals
were assessed separately for each phase and each score.
Different grading systems were compared by using the Bayesian Information
Criterion (BIC) calculated for multivariate regression analysis :
1. BIC difference : 2–6, positive
2. BIC difference : 6 –10, strong
3. BIC difference : >10, very strong
63
Figure 59. Examples of the volume of hypoattenuation and collateral vessels. A and B, Examples of 2 typical
patients. The top rows show 10-mm MIPs of each phase in dynamic CTA. The areas of hypoattenuation of
these MIPs are outlined, and the total lesion volumes of each phase are given beneath the images. The bottom
rows show 20-mm MIPs of each phase in dynamic CTA, illustrating the collateral vessels. Collateral grades for
each phase are given beneath the images and correspond to (in the order given) the 20-point relative score, the
5-point relative score, and the 4-point absolute score. The area of infarction on follow-up is shown on the right.
Reformatted images of the carotid and vertebral arteries are obtained for better
evaluation of the presence and morphologic features (calcification, irregular
surface, ulceration) of heterogeneous plaques and quantification of the degree of
stenosis
Figure 60.
64
A whole-brain analysis of the source images with a narrow window provides a
whole-brain “perfused blood volume map,” since the contrast agent fills the brain
microvasculature in normal perfused tissue but not the microvasculature of
infarcted brain regions, depicted as hypoattenuating parenchyma
Figure 61.
3.3.4. Characterizes carotid atherosclerotic disease (Figures 62,63 and Table 13)
65
CT angiography–source imaging is more sensitive than nonenhanced CT in the
detection of early irreversible ischemia and more accurate in predicting final
infarct volume, with good correlation with the hyperintense lesions seen at
diffusion-weighted imaging and the low-CBV areas seen at perfusion CT. It
holds for a sensitivity : (48% - nonenhanced) and (70% CT Angio - Source
Images) and a specificity : 100% for both.
CT angiography–source imaging provides accurate complementary whole-brain
information for the perfusion CT maps and can sometimes obviate a separate
perfusion CT study.
3.4. Brain CT Perfusion (CTP)
3.4.1. CT Perfusion general principles
CT Perfusion is performed in cine mode (repeated image acquisition at one couch
position), at the level of the basal ganglia because it contains representative territories
of the Anterior Cerebral Artery (ACA), Middle Cerebral Artery (MCA) and Posterior
Cerebral Artery (PCA), which is the best option for evaluating suspected MCA stroke.
However, one can also set the region elsewhere although many more artifacts might be
observed.
66
Figure 64. Figure 65.
In acute stroke, there is an irreversibly infarcted tissue core surrounded by a peripheral
region of stunned cells called the penumbra that receives a collateral blood supply from
uninjured arterial and leptomeningeal territories
Figure 66.
Dynamic Contrast Enhanced CT
Discriminates Penumbra (salvageable tissue) / Infarct Core (necrotic)
Figure 67.
67
CT Perfusion achieves the discrimination of the penumbra from the infarct core by
calculating the following three quantitative parameters :
1. Mean Transit Time (MTT) : MTT represents the average time required for a
blood passage through the capillary network.
2. Cerebral Blood Volume (CBV) : CBV holds for the total blood volume within a
given volume of cerebral parenchyma.
3. Cerebral Blood Flow (CBF) : CBF is the volume of blood perfusing a given
volume of brain parenchyma, per unit time.
Of the three parameters presented above, MTT should be analyzed first because it
shows the most prominent regional abnormalities and facilitates depiction of the
ischemic area (penumbra) and the search for a correlation with suspect clinical and
imaging findings. Subsequently, analysis of the CBF and CBV maps is conducted, since
these are more specific for distinguishing ischemia from infarction.
CT perfusion tracks the first pass of a bolus of contrast by acquiring a rapid
series of images without table movement following Intra Venous contrast
injection and generates maps of cerebral blood flow (CBF), cerebral blood
volume (CBV) and mean transit time (MTT)
Figure 68.
68
3.4.2. Quantitative Analysis in CT Perfusion
Delineation of the salvageable brain tissue (penumbra) with :
1. Increased Mean Transit Time (MTT)
2. Normal or Increased Cerebral Blood Volume (CBV)
3. Decreased Cerebral Blood Flow (CBF)
Table 14.
Manifestation of infarcted tissue with :
4. Increased Mean Transit Time (MTT)
5. Markedly decreased Cerebral Blood Volume (CBV)
6. Markedly decreased Cerebral Blood Flow (CBF)
Table 15.
69
Overall, MTT, CBF and CBV are summarized for the penumbra and the infarct core
(Table 16):
Suggested quantitative thresholds for the MTT and CBV are presented below
(Table 17):
3.4.3. Qualitative Analysis in CT Perfusion
The CBV map depicts the infarcted brain tissue that is not salvageable
The CBF map depicts the ischemic area
The salvageable brain tissue is equivalent to CBF – CBV
-55% -82% -28% -75% 130% 110%
70
Percentage of mismatch (CBV area / CBF area)
Figure 69.
Figure 70.
71
Figure 71.
3.4.4. Calculation of the specific parameters : MTT, CBV, CBF
CT Perfusion requires continuous cine imaging over the same slab of tissue during
the dynamic administration of a small quantity of high-flow contrast material bolus
(injection rate : 4–5 mL/sec). Therefore, it is important to detect possible motion artifact
prior to selecting the region of interest (ROI) vessels. The contrast agent causes a
transient hyperattenuation directly proportional to the amount of contrast material in
the vessels and blood in that region. The dynamic first-pass approach to CTP
measurement involves the intravenous administration of an intravascular contrast agent,
which is tracked with serial imaging during its first circulation through the brain tissue
capillary bed. The main assumption of dynamic first-pass contrast-enhanced CTP
models is that the perfusion tracer is not diffusible, neither metabolized nor absorbed
by the tissue through which it traverses. This is certainly the case in a healthy human
72
brain; however, breakdown of the blood-brain barrier (BBB) in infection,
inflammation, or tumor adds an additional level of complexity. When extensive BBB
breakdown exists, leakage of contrast material into the extravascular space results in
overestimation of CT CBV.
The determination of cerebral perfusion by using CTP is based on examining the
relationships between the arterial, tissue, and the venous enhancement. More
specifically, tracer kinetic theory states that if one knows the input and the output of a
tracer from a voxel, one can determine the volume of distribution (i.e., fractional
vascular volume) and the clearance rate (i.e., flow per unit tissue volume). The
fractional vascular volume, f, is defined by the following equation:
(27)
where Vvasc , Vinterstitium , and Vcells are the volumes occupied by the vascular space,
interstitium, and cells, respectively. If the chosen region of interest is devoid of major
blood vessels, the measured change in the CT number will reflect the tissue blood pool.
The contrast concentration in the tissue, Ctissue, which is measured by the CT scanner,
is smaller than the intravascular concentration, Cvasc, by the fraction f,
(28)
The total amount of contrast material delivered to the tissue via the arteries is the
product of the CBF times the integral of the arterial concentration, Cartery(t). According
to the conservation-of-mass principle, this total amount must be equal to the amount
leaving the tissue, that is, the product of CBF with the integral of Cvasc(t). Hence,
(29)
From equations 28 and 29, it follows that:
(30)
73
CBV can be calculated from equation 31 if one takes into account the brain tissue
attenuation, ρ , and a correction factor, CH, to adjust the difference between arterial and
capillary hematocrit. In vivo experiments have demonstrated the markedly lower
hematocrit in capillaries compared with arterial hematocrit; hence, the introduction of
the CH is required for the quantification of CBV as follows:
(31)
Theoretic modeling has suggested that the source images from a CTA acquisition
(CTA-SI) are predominantly blood-volume (rather than blood-flow) weighted,
assuming that a steady state of arterial and tissue contrast has been achieved during
scan acquisition. Coregistration and subtraction of the unenhanced head CT images
from the CTA-SI should, therefore, result in quantitative CBV-weighted maps. This is
appealing in clinical practice because CTA-SI subtraction maps, unlike first-pass CTP
maps, can provide whole-brain coverage. The change in attenuation due to iodine
administration is directly proportional to its concentration; thus, the ratio of the change
in Hounsfield units (HU) in brain tissue (ΔHUtissue) and arterial blood (ΔHUartery) can
be used in practice to estimate CBV according to equation 32 (Fig. 72).
Fig. 72. CTA source images acquired during a steady state of contrast concentration for both
the arterial and tissue time-attenuation curves (ΔT) are predominantly blood-volume (rather
than blood-flow) weighted. The change in attenuation due to iodine administration is directly
proportional to its concentration. CBV equals the ratio of the areas under the 2 curves, Ctissue
and Carterial, respectively. This can be approximated as the ratio of the HUtissue/HUarterial when the
2 curves approach steady state.
74
Specifically, CBV can be expressed in milliliters per 100 g of tissue as follows:
(32)
where Vvoxel is voxel volume and N is the calculated number of voxels in 100 g of tissue.
Although a relative steady state of tissue arterial contrast could be assumed with the
multidetector row CT (MDCT) injection protocols used in which low contrast-injection
rates were applied with relatively long prep delay times (typically 3 mL/s, and >25
seconds), this steady-state assumption no longer holds in the era of newer and faster
MDCT CTA protocols, such as those used in 64-section scanners, in which injection
rates ≤7 mL/s and short prep delay times of 15–20 seconds change the temporal shape
of the time-attenuation infusion curve, eliminating a near steady state during the timing
of the CTA-SI acquisition. Hence, with the current implementation of CTA protocols
on faster state-of-the-art MDCT scanners, the CTA-SI maps have become more flow-
weighted than volume-weighted.
The 2 major mathematic approaches involved in calculating CBF and MTT are the
deconvolution and nondeconvolution based methods. Deconvolution techniques are
technically more demanding and involve more complicated and time-consuming
processing, whereas nondeconvolution techniques are more straightforward but depend
on simplified assumptions regarding the underlying vascular architecture. As a result,
the interpretation of studies based on nondeconvolution methods may be less reliable
in some situations, though this has not been fully clinically validated.
Nondeconvolution-Based Models
Like deconvolution methods, nondeconvolution methods are based on the application
of the Fick principle of conservation of mass to a given region of interest within the
brain parenchyma (Fig. 73). The accumulated mass of contrast, Q(T), in a voxel of brain
tissue during a time period corresponding to the complete wash in and wash out of
contrast during a bolus injection (with a saline “chaser”) is equal to the product of CBF
and the time integral of the arteriovenous difference in contrast concentration:
(33)
75
Figure 73.
One immediate simplification that would make solving this equation less
computationally intensive is to assume that during the above time period, the venous
concentration is zero (i.e., no contrast has yet reached the venous side of the circulation,
the “no venous outflow” assumption). This assumption is valid only when T is <4–6
seconds, the minimum transit time of blood through the brain. Under this assumption,
equation 7 can be simplified as follows:
(34)
This is known as the Mullani-Gould formulation or single-compartment formulation.
When rewriting equation 34 into its derivative form for easier calculation of CBF,
(35)
it follows that the rate of contrast accumulation will peak when the arterial
concentration is maximal:
(36)
76
Hence, CBF is the ratio of the maximum slope of Q(t) to the maximum arterial
concentration. This is analogous to a differentiation with respect to time of the Mullani-
Gould formulation and is called the “maximum slope method”. The principal advantage
of the maximum slope method is the simplicity and hence speed of calculation of the
perfusion values. A very high rate of contrast agent injection, however, is required
(typically at least 10 mL/s) to satisfy the no-venous- outflow assumption. These rates
cannot be routinely achieved in clinical practice. The no-venous-outflow assumption is
clearly an oversimplification, and this method yields relative, rather than absolute,
perfusion measurements, making interpatient or interinstitutional comparison of results
difficult.
Deconvolution-Based Models
Considering a bolus of contrast tracer material injected into a tissue voxel of interest,
we can define the concentration Ctissue(t) of tracer in the tissue in terms of 2 functions:
1) Residue function, R(t): a fraction of tracer is still present in the voxel of interest at
time t following an ideal instantaneous unit bolus injection. R(t) is unitless and is equal
to 1 at t=0;
2) AIF, Cartery(t): concentration of tracer in the feeding vessel to the voxel of interest at
time t.
Calculation of CBF requires measurement of the temporal shape of both the arterial
input and the tissue time-attenuation curves. The true input into the tissue voxel of
interest cannot be measured directly; in practice, the AIF is estimated from a major
artery (e.g. the MCA or the “top” of the internal carotid artery ICA), with the
assumption that this represents the exact input to the tissue of interest. Any delay or
dispersion of the bolus introduced during its passage from the artery used for AIF
estimation to the tissue of interest will introduce errors in quantification of CBF. The
observed tissue time-attenuation curve represents a combination of the effects of the
AIF and the inherent tissue properties. Thus, to fit the model, the effects of the AIF on
the tissue concentration curve must be removed by using a mathematic process known
as “deconvolution” to derive R(t), which is dependent only on the hemodynamic
properties of the voxel under consideration. R(t) demonstrates an abrupt (indeed
77
instantaneous) rise, a plateau of duration equal to the minimum transit time through the
tissue of interest and then decay toward baseline. It has shown that the tissue
concentration curve can be represented as the CBF multiplied by AIF convolved with
R(t):
(37)
where is the convolution operator, ρ is the brain tissue attenuation, and CH is the
correction factor for the capillary hematocrit levels. Ctissue and AIF are measured
directly from the time-attenuation curve from the cine CTP source images, and the
problem then becomes the calculation of CBF and R(t). Several methods to
“deconvolve” the previous equation and hence solve for CBF and R(t) have been
proposed and are divided into 2 main categories. With parametric methods, a specific
analytic expression for R(t) is assumed. Assuming a specific shape for R(t) imposes
assumptions on the inherent tissue properties that cannot be known a priori with
sufficient precision. Due to this limitation, the most commonly used methods have
become the nonparametric ones, which do not assume a shape for R(t). Deconvolution
of the previous equation is unstable, in the sense that it frequently yields nonphysiologic
oscillations (i.e., noise) in the computation of the solution for R(t). The nonparametric
approach can be further subdivided in 2 categories, which differ in the way in which
they deal with noise resulting from the instability of deconvolution. In the first, the
transform approach, the convolution theorem of the FT is used to deconvolve the
previous equation. The theorem states that the FT of the convolution of 2 time-domain
functions is equivalent to the multiplication of their respective FT’s. With the
convolution theorem, equation 11 can be rewritten as:
(38)
where ℑ is the Fourier transform. R(t) and CBF can thus be determined by taking the
inverse FT of the ratios of the 2 transforms of the sampled AIF(t) and Ctissue(t).
However, this approach is very sensitive to noise. The second approach, the more
commonly applied singular value decomposition (SVD), is an algebraic reformulation
of the convolution integrals of equation 37 rewritten as :
78
(39)
where t1, t2,…tN are equally spaced time points and AIF(t) and Ctissue(t) are measured.
Equation 39 can be rewritten in shorthand matrix-vector notation:
A ∙ b = c (40)
where b and c are vectors whose elements are R(ti) and Ctissue(ti), respectively, and A is
the convolution matrix in equation 39. It can be shown that the least squares solution
for b is given by (AT ∙A)-1 ∙ AT, where AT is the transpose of the convolution matrix and
(AT ∙A)-1 is the inverse of the symmetric matrix AT ∙A. SVD decomposes A into a
product of matrices, such that (AT ∙A)-1 can be found easily and the solution for b can
be written as a sum of terms weighted by the reciprocal of the singular values of A. By
truncating small singular values in the sum for the solution vector b, oscillations from
noise in the both AIF(t) and Ctissue(t) are avoided. SVD has yielded the most robust
results from all the deconvolution methods used to map CBF and has gained widespread
acceptance. The creation of accurate quantitative maps of CBF, CBV, and MTT by
using deconvolution methods has been validated in a number of studies. Once CBF and
CBV are known, MTT can be calculated from the ratio of CBV and CBF, by using the
central volume theorem. Potential pitfalls in the computation of CBF by using
deconvolution methods include patient motion and partial volume averaging, which can
cause underestimation of the AIF(t). Image-coregistration software to correct for patient
motion and careful choice of regions of interest for the AIF can minimize these pitfalls.
Commercial software suppliers use different mathematic methods. In the past, some
have incorporated the maximum slope method, though new versions are frequently
released and the reader is advised to check for the most up-to-date software from each
vendor. Others have typically used deconvolution techniques, which, though
79
theoretically superior to nondeconvolution methods, the full clinical implication of
using has yet to be established and standardized by the stroke imaging community.
Validation of CTP
CBF quantification with CTP has been preliminarily tested in humans. The accepted
reference standard method for quantifying CBF is the microsphere technique and, to a
lesser extent, postmortem histology. CTP measurements have been validated in
ischemic stroke, and tumors. All studies reported very good correlation between CTP
and the reference-standard methods. CTP has been validated with positron-emission
tomography and xenon-enhanced CT in healthy subjects, patients with acute stroke, and
patients with chronic cerebrovascular disease. Deconvolution-based CTP studies by
gave slopes within 13% of unity, whereas the difference from unity was 20%–60% for
the maximum slope technique. These results once again suggest a superior accuracy of
deconvolution techniques compared with maximum slope based methods.
The principle of transient hyperattenuation is used to generate time-attenuation curves
for an arterial ROI, a venous ROI, and each pixel.
Figure 74.
80
The Arterial ROI is optimally selected in one unaffected vessel perpendicular to the
acquisition plane, either one of the anterior cerebral arteries (ACAs) or the
contralateral MCA
Figure 75.
The Venous ROI is placed over the superior sagittal sinus or torcular Herophili
(adequate window width must be used so as not to include skull bone within the
venous ROI).
Figure 76.
The venous ROI is necessary to correct the data for partial volume averaging
effects to help achieve accurate quantification of perfusion parameters
81
Partial volume artifact occurs when tissues of widely different absorption are
encompassed on the same CT voxel producing a beam attenuation proportional
to the average value of these tissues. Last generation CT scanners reduction of
the voxel volume has substantially reduced this artifact’s occurrence.
Thus, the generated time-attenuation curve must be studied :
1. To detect possible poor timing of the contrast material bolus
2. To distinguish good arterial input function or venous outflow function MTT
Figure 77.
MTT is calculated by deconvolution on the regional time-attenuation curve of each
pixel with respect to the arterial curve (arterial input function)
Figure 78.
82
CBV is calculated by dividing the area under the curve in a parenchymal pixel by
the area under the curve in an arterial pixel
Figure 79.
The quantification of these parameters is based on the equation CBF = CBV/MTT
and is estimated as follows
Figure 80. See References 4,5 and 6.
83
3.4.5. Multimodal CT Evaluation Aim
1. To improve detection of acute infarction
2. Permit assessment of the site of vascular occlusion
3. Discriminate the infarct core and salvageable brain tissue
4. Help assess the degree of collateral circulation
5. Be performed and analyzed rapidly and easily by general radiologists using a
simple standardized protocol
6. Correlation of all the imaging findings with the vascular anatomy and clinical
findings is crucial
7. CT acute stroke imaging provides important anatomic and physiologic data for
treatment decisions
8. Radiation doses must be balanced with the benefits to the patient
9. CT acute stroke protocols should be monitored by the radiologist and tailored
to the patient
84
Chapter 4. CT Perfusion as a diagnostic tool
4.1. CT Perfusion in acute stroke
Brain CT Perfusion (CTP) is an X-ray imaging technique for the assessment of
various cerebrovascular disorders, especially in acute stroke patients[1-3]. In spite of the
fact that Magnetic Resonance (MR) perfusion-diffusion weighted imaging have been
settled in clinical practice, MRI is mostly unavailable in non stroke-dedicated hospitals
and inaccessible to patients with metallic implants or a pacemaker[4,5]. Stroke is
attributed to perturbation of blood flow in the brain, due to either obstruction (ischemia)
or fissure (hemorrhage) of a blood vessel[5].
4.2. Evaluation and Anatomical Region
Hemorrhage may be precluded by performing an unenhanced brain CT (UN-CT) scan,
followed by a brain CT Angiography (CTA) to reveal the occlusion presence and site,
whilst a CTP scan allows the distinction of the potentially salvageable ischemic
penumbra from the infarct core(5-8). The discrimination of potentially salvageable
ischemic penumbra from the infarct core, is accomplished by capturing the first pass of
a bolus of contrast medium through the vasculature, a procedure that requires
consecutive imaging of the identical parenchymatic region[7-9]. This area lies at the
plane of the basal ganglia as it comprises vicarious territories of the Anterior, Middle
and Posterior Cerebral Artery, offering the best spatial selection for stroke
assessment[5,10].
4.3. Associated Ionizing Radiation
Nevertheless, exposure to X-ray radiation, associated with the consecutive imaging of
the CTP, combined with the additional exposure from UN-CT and CTA, and the fact
that stroke patients may be submitted to manifold proceedings, declares that the
management of the patient X-ray dose commensurate with the medical task of the
justification of the procedure, is a necessity[3,10,11]. Numerous studies have reported high
85
dose results, suggesting the need for further investigation[10-14,16]. The most useful dose
quantity used in these studies is the Effective Dose (ED), as it facilitates the comparison
amongst different techniques[17,18]. The ED derived from the following equation : ED =
k x DLP (41), where k is an appropriately selected conversion factor[4,10,13,18].
4.4. Aim of the study
To our knowledge, this is the first study that aims to present radiation dose to patients
during a comprehensive brain CT prescription protocol (CPP), carried out in our
hospital using an 80-slice CT system and an iterative reconstruction (IR) algorithm.
Chapter 5. CT Perfusion in clinical practice
5.1. Patient data and utilized CT system
Fifteen patients participated in this study undertaking brain CT imaging for either
acute stroke, brain death verification, vasospasm following SAH, or brain tumor.
Patients’ data such as sex, age and weight were recorded. Patients included 8 women
and 7 men. The mean age of all patients was 51.8 y, ranging between 18 and 78 y. The
mean weight of all patients was 78.1 kg, ranging between 60 and 100 kg.
All the brain CT procedures were carried out at the Department of Radiology, at the
University Hospital of Patras, between January and June 2017. The system used was an
80-slice CT scanner (Toshiba Aquilion Prime, Toshiba Medical Systems Corporation),
with a maximum beam collimation of 40 mm (e.g. 80 x 0.5 mm). The utilized algorithm
for the specific study was the IR algorithm AIDR 3D Std.
5.2. Protocol and Contrast Agent parameters
The comprehensive prescription protocol (CPP) consists of three types of
examinations; an UN-CT, a CTP and a CTA at a tube voltage of 120, 80 and 120 kVp,
respectively. The scan range for the UN-CT and CTA was from the skull base to vertex
performed in helical scan mode. CTP scanning was performed with the maximum beam
86
collimation of 40 mm (10 x 4 mm), in axial mode, with the center of the scan length
being posed at the plane of the inferior part of the bodies of lateral ventricles, during
the dynamic administration of 50 ml of iodinated contrast media with a concentration
of 350 - 370 mg/ml of iodine, intravenously injected at a flow rate of 5 ml/s, followed
by 30 ml saline at 5 ml/s. Additionally, CTA scanning was performed with the dynamic
administration of 100 ml of iodinated contrast media with a concentration of 350 - 370
mg/ml of iodine, intravenously injected at a flow rate of 4 ml/s, followed by 30 ml
saline at 4 ml/s.
Overall, the scan parameters for the three types of examinations of the CPP are
presented in Table 18. The system was under a periodic quality control program, in
order to ensure the correct performance of the equipment, as well as the reliability and
reproducibility of the exposure parameters.
5.3. Dose report data
The dose report of the system provided data regarding the tube voltage (kVp), Volume
Computed Tomography Dose Index (CTDIvol) and DLP. CTDIvol represents the average
absorbed radiation dose within the irradiated scan volume, for a similar attenuation
standardized phantom, while DLP is the product of CTDIvol and the length of the
imaged object.
5.4. Conversion Factor for the estimation of the Effective Dose
A reasonable approximation for the patient ED can be obtained using the conversion
factor of 0.0021 mSv/mGy∙cm, since conversion factor values in bibliography, vary
from 0.0019 to 0.0023 mSv/mGy∙cm. The conversion factor value used, is applicable
to head scans. The patient ED was thus calculated by multiplying each DLP value with
the conversion factor of 0.0021 mSv/mGy∙cm[4,14,18].
87
Table 18. Scan parameters for the CPP of this study.
Modality UN-CT CTA CTP
Tube Voltage (kVp) 120 120 80
Tube Current (mA) 230 250 40
Rotation Time (s) 0.75 0.5 1.5
Pitch Factor 0.625 0.637 1
Scan Type Helical Helical Axial
Anatomical Area Scull base to vertex Scull base to vertex Basal ganglia
Beam Collimation (mm) 40 x 0.5 mm 80 x 0.5 mm 10 x 4 mm
Contrast Agent None 100 ml of 350 - 370 mg/ml
Iodine contrast at 4 ml/s followed
by 30 ml saline at 4 ml/s
50 ml of 350 - 370 mg/ml Iodine
contrast at 5 ml/s followed by 30
ml saline at 5 ml/s
88
Chapter 6. Dosimetry Results
6.1. CTDIvol for the comprehensive prescription protocol
The mean CTDIvol for the UN-CT was 55.4 mGy, ranging from 54.7 to 60.1 mGy.
The mean CTDIvol for the CTA was 35.2 mGy, ranging from 34.7 to 38.2 mGy. The
mean CTDIvol for the CTP was 119.8 mGy, ranging from 116 to 134.8 mGy. The results
are depicted in Figure 81.
6.2. DLP for the comprehensive prescription protocol
Furthermore, the mean DLP for the UN-CT was 912.1 mGy∙cm, ranging from 473.7
to 1080.9 mGy∙cm. The mean DLP for the CTA was 678.6 mGy∙cm, ranging from
543.7 to 797.9 mGy∙cm. The mean DLP for the CTP was 478.9 mGy∙cm, ranging from
463.9 to 539.2 mGy∙cm. The results are presented in Figure 82.
89
6.3. Effective Dose for the comprehensive prescription protocol
Finally, the mean ED value for the UN-CT was 1.91 mSv, ranging from 0.99 to 2.27
mSv. The mean ED for the CTA was 1.42 mSv, ranging from 1.14 to 1.68 mSv. The
mean ED for the CTP was 1 mSv, ranging from 0.97 to 1.13 mSv, whilst the mean ED
for the CPP was 4.34 mSv, ranging from 3.31 to 4.88 mSv. The results are showed in
Figure 83. The values for the CTDIvol, DLP and ED are presented in Table 19.
90
Table 19. CTDIvol, DLP and ED values for the CPP of this study.
CTDIvol (mGy) DLP (mGy∙cm) ED (mSv)
UN-CT CTA CTP UN-CT CTA CTP UN-CT CTA CTP CPP
Min 54.7 34.7 116.0 473.7 543.7 463.9 0.99 1.14 0.97 3.31
Max 60.1 38.2 134.8 1080.9 797.9 539.2 2.27 1.68 1.13 4.88
Mean 55.4 35.2 119.8 912.1 678.6 478.9 1.91 1.42 1.00 4.34
SD (±) 1.89 1.22 7.78 137.98 60.14 31.17 0.29 0.13 0.07 0.37
CV (%) 3.41 3.46 6.49 15.13 8.86 6.50 15.18 9.15 7.0 8.52
91
Chapter 7. Discussion and Conclusion
The results of the present study were compared with the corresponding values
reported in the literature and are presented in Table 20 for the UN-CT, CTA and CTP,
in terms of CTDIvol, DLP and ED.
7.1. Comparison of CTDIvol with the literature
In particular, for CTP, the mean CTDIvol in our study was 119.8 mGy. Arandjic et
al.[11], Hoang et al.[12] and Bricout et al.[14] using 64-slice CT systems, and a tube load
of 200, 200 and 180 mAs, respectively, reported higher CTDIvol values, than the mean
value of our study. Moreover, Horiguchi et al.[13] also using a 64-slice CT system and
a tube load of 80 mAs, reported a comparable CTDIvol value to the mean value of our
study. Fang et al.[16], utilising a 128-slice CT system and a tube load of 100 mAs,
without referring any other protocol parameters, reported a lower CTDIvol value than
the mean value of our study. Li et al.[20] and Solano et al.[21] utilising 128-slice CT
systems as well, and a tube load of 150 and 80 mAs, respectively, reported higher
CTDIvol values, than the mean value of our study. Brix et al.[22] using a 192-slice CT
system and a tube load of 250 mAs, reported a CTDIvol value of 196 mGy which is 63%
higher than our value, although they used a tube voltage of 70 kVp. Lin et al.[23] using
a 256-slice system and a tube load of 80 mAs, reported a CTDIvol value of 128.2 mGy,
which is 7% higher than our value. Salomon et al.[32] using a 320-slice system and a
variable tube load, reported a range of CTDIvol values, with the minimum value of 400
mGy, which is 3.34 fold higher than our value.
7.2. Comparison of DLP with the literature
Regarding the DLP, in CTP, the mean value in our study was 478.9 mGy∙cm.
Mnyusiwalla et al.[4], Arandjic et al.[11], Hoang et al.[12], Horiguchi et al.[13] and Bricout
et al.[14], using 64-slice CT systems, reported values 2-5.6 fold higher than the mean
value of our study. Mnyusiwalla et al.[4], performed CTP in 190 mAs compared to the
value of 60 mAs of our CTP exam, whereas Arandjic et al.[11], conducted CTP in 200
mAs and a scan length of 8 cm, compared to 4 cm in our study. In addition, Hoang et
92
al.[12], Horiguchi et al.[13], and Bricout et al.[14], performed CTP in 200, 80, and 200
mAs, respectively, with Horiguchi et al.[13], and Bricout et al.[14], conducting the exam
in an extended 8 cm coverage. Fang et al.[16], Li et al.[20] and Solano et al.[21], utilising
128-slice CT systems, reported values 1.8-4.5 fold higher than our value as these studies
performed CTP in an extended coverage of 15 cm, 12 cm, and 12cm, respectively. Brix
et al.[22], using an 192-slice CT system, reported a DLP value of 2909 mGy∙cm, which
is 6.1 fold higher than our value, as their scan length was 11.4 cm. Lin et al.[23] and
Dorn et al.[25], using 256-slice CT systems, reported DLP values 1.5-2.1 fold higher, as
Lin et al.[23], had an extended 8 cm coverage and Dorn et al.[25], performed CTP in 120
mAs and an 8 cm coverage, compared to 60 mAs and 4 cm z-axis coverage in the
present study. Shankar et al.[26] using 100 mAs and a 16 cm z-axis coverage and
Salomon et al.[30], without providing data on the scan length, with both studies using
320-slice CT systems, reported DLP values 2.1-4.7 fold higher than our value.
7.3. Comparison of Effective Dose with the literature
For the calculation of the ED there are two main methodologies; the first is by
multiplying the DLP values with an appropriately selected conversion factor[4,10,13,18],
and the second is by adding measured organ doses as that recorded by
thermoluminescent dosimeters placed in an anthropomorphic phantom and multiplied
by weighting factors which are provided by international organizations (ICRP
Publication 103)[31]. Conversion factor values vary from 0.0021-0.0023
mSv/mGy∙cm[4,14,16,18,21,28,30], whereas we used the conversion factor of 0.0021
mSv/mGy∙cm.
Regarding the ED, in CTP, the mean value of our study was 1 mSv. Mnyusiwalla et
al.[4], Sabarudin et al.[10], Arandjic et al.[11], Hoang et al.[12], Horiguchi et al.[13], Bricout
et al.[14] and Konstas et al.[15], using 64-slice CT systems, reported values 2.1-7.5 fold
higher. Mnyusiwalla et al.[4], had a value of 4.90 mSv, which is almost five times higher
than our mean ED value of 1 mSv, derived from their large mean DLP and the same
conversion factor we also used in our study. Moreover, our value was lower than
Sabarudin et al.[10], despite the fact that their presented ED was estimated in another
way than our study, by placing dosimeters in an anthropomorphic phantom, as their
protocol was conducted in 80 kVp, 270 mAs and a scan length of 5 cm, while from the
recorded DLP values and the conversion factor of 0.0022 mSv/mGy∙cm, they came to
93
the reported ED value. Hoang et al.[12], presented their ED as an estimation from the
measured average organ doses recorded by dosimeters placed in an anthropomorphic
phantom. Arandjic et al.[11] and Horiguchi et al.[13], indicated a larger ED value, as their
higher DLP value, was multiplied by the greater than ours, conversion factor of 0.0023
mSv/mGy∙cm, although not mentioned in the latter case. Bricout et al.[14], utilized the
same conversion factor as in the present study, to calculate the higher than ours, ED of
3.90 mSv. Konstas et al.[15], in a CTP protocol performed in 80 kVp, 200 mAs and 4
cm z-axis coverage, estimated an ED of 3.70 mSv, which is 4 times higher than ours.
Fang et al.[16], Saake et al.[19], Li et al.[20] and Solano et al.[21], utilising 128-slice CT
systems, reported values 1.8-4.7 fold higher. Fang et al.[16], used the conversion factor
of 0.0023 mSv/mGy∙cm, while Saake et al.[19], presented their ED after personal
communication. Moreover, Li et al.[20], does not mention the manner in which they
calculated their ED value, whereas, Solano et al.[21], used the conversion factor of
0.0023 mSv/mGy∙cm. Brix et al.[22] using an 192-slice CT system and the conversion
factor of 0.0022 mSv/mGy∙cm, reported a value of 6.4 mSv, which is 6.4 fold higher
than our value.
Lin et al.[23], Murayama et al.[24] and Dorn et al.[25], using 256-slice CT systems, reported
values 1.7-3.5 fold higher. Lin et al.[23], estimated the ED using the conversion factor
of 0.0021 mSv/mGy∙cm, although not mentioned. Murayama et al.[24], calculated a
minimum ED of 3.50 mSv, in a CTP protocol conducted in 80 kVp, 80 mAs and a
scanning range of 12.8 cm, while Dorn et al.[25], presented an ED 1.67 mSv, 67% higher
than our value of 1 mSv, using the conversion factor of 0.0023 mSv/mGy∙cm, although
not mentioned. Manniesing et al.[27], Orrison et al.[28], Diekmann et al.[29] and Salomon
et al.[30], using 320-slice CT systems, reported values 3.6-5.1 fold higher than our value.
Manniesing et al.[27], performed a CTP protocol in 80 kVp, with a tube load modulation
starting at 200 mAs lowering to 100 and ultimately to 75 mAs, and a 16 cm coverage,
to indicate an ED of 5 mSv. Orrison et al.[28], using a tube load modulation (300 mAs
to 100 mAs) a z-axis coverage of 16 cm and the conversion factor of 0.0023
mSv/mGy∙cm presented a value 4.3 fold higher than our value. Diekmann et al.[29],
presented their ED value of 3.61 mSv as an estimation from the measured average organ
doses recorded by dosimeters placed in an anthropomorphic phantom. Salomon et
al.[30], using a tube load of 100 mAs and the conversion factor of 0.0023 mSv/mGy∙cm
presented a minimum value of 5.13 mSv.
94
Overall, the ED values reported in the literature along with the mean ED of our study
for CTP, in comparison with the number of slices of the CT systems used, are presented
in Figure 84. According to this figure, it is revealed that the 80-slice CT system of our
study has a mean ED that is lower than that of 64-slice CT systems, as well as that of
128,192,256 and 320-slice CT systems.
7.4. Comparison of Total Effective Dose with the literature
For the CPP, the total ED value of 4.34 mSv of this study was compared only with
these studies from the literature that included at least the three types of examinations
(UN-CT, CTA, CTP). Mnyusiwalla et al.[4], Bricout et al.[14], Saake et al.[19] and
Manniesing et al.[27] reported values 1.4-2.9 fold higher than our value, mainly due to
the larger dose contribution of their CTP protocols in total ED (see Table 20). At this
point, it should be noted that, the value reported by Mnyusiwalla et al.[4], corresponded
to two additional examinations; CT permeability and post contrast CT, whereas Bricout
et al.[14], reported median values.
7.5. Comparison of the protocol Effective Dose with Background Radiation in
Greece
Studies performed by the Greek Atomic Energy Commission have showed that the
amount of background radiation received by people living in the country is 2.7 mSv per
year[32]. Therefore, for our study the mean ED for the UN-CT of 1.91 mSv is equal to
8.5 months of background radiation. The mean ED for the CTA of 1.42 mSv is equal
to 6.3 months of background radiation. The mean ED for the CTP of 1 mSv is equal to
4.5 months of background radiation. Finally, the mean ED for the CPP of 4.34 mSv is
equal to 19 months of background radiation. The results are depicted in Figure 85.
95
7.6. Advantages and Limitations of the present study
To our knowledge this is the first study that referred to a CPP for brain perfusion
using an 80-slice CT system. According to this, the radiation dose to the patients was
found to be significantly lower compared to the values previously reported using CT
systems with slices less or more than that of our 80-slice CT system. Regarding the
characteristics of the present study, the main limitation of our study was the small group
of patients participated (N = 15).
7.7. Future work
Since the CTP protocol conducted in our institution presents a small Standard
Deviation value (± 0.07 mSv), any changes shall arise from the optimisation of the UN-
CT and CTA, which constitutes future work. The values presented in this study can
contribute in the effort for the establishment of national Diagnostic Reference Level
(DRL) values.
96
Overall, using modern multislice CT systems in brain CTP examinations, an increase
of slices per rotation from 64 to 320, does not mandatorily result in a decrease in patient
dose. It appears that the 80-slice CT system versus the commercially available CT
scanners varying from 64 to 320 slices, holds for the best comparative result. Thus, the
collaboration between the Radiologist and the Medical Radiation Physicist of each
institution is of paramount importance for the adoption of an optimum prescription
protocol that will secure the fidelity of the examination from the medical point of view
and simultaneously minimize the radiation dose to the patient by ensuring that the
ALARA principle is in effect.
7.8. Conclusion
In this study, the Effective Dose concerning a comprehensive prescription protocol
about the brain CT Perfusion imaging modality was presented and compared with the
existing bibliography. The comparison reveals that the reduction of radiation dose to
the patient is not a priori lower using CT systems of increased slices. On the contrary,
it appears that for the time being, the 80-slice CT system provides ED’s that are lower
than that of a 64,128,192,256 and 320 system slices. Thus, radiation dose to patients
was found to be at least 67% lower compared to the values previously reported in
studies utilizing other CT systems. The results highlight the necessity and importance
of the harmonic collaboration between the radiologists and the medical physicists,
leading to the enhanced utilisation of the system used.
97
Table 20. Comparison of CTDIvol, DLP and ED mean values in UN-CT, CTA and CTP reported in various studies.
Study
System
Slices
CTDIvol (mGy) DLP (mGy∙cm) ED (mSv)
UN-CT CTA CTP UN-CT CTA CTP UN-CT CTA CTP Total
Loftus et al.[3] - - - - 1183.2 3128.2 1991.8 - - - -
Mnyusiwalla et al.[4] 64 - - - 1227.8 1565.4 2663.6 2.70 1.60 4.90 12.5c
Sabarudin et al.[10] 64 - - - - - - - 0.61 2.07 2.68
Arandjic et al.[11] 64 - - 230.0 - - 2120.0 - - 4.90 4.90
Hoang et al.[12] 64 - - 531.4 - - 2125.7 - - 7.50 7.50
Horiguchi et al.[13] 64 - - 121.0 - - 965.0 - - 2.22 2.22
Bricout et al.[14] 64 42.6a 24.5a 204.1a 662.0a 369.0a 1831.0a 1.40a 0.80a 3.90a 6.10a
Konstas et al.[15] 64 - - - - - - - - 3.70 3.70
Fang et al.[16] 128 - - 55.5 - - 859.0 - - 1.80 1.80
Saake et al.[19] 128 - - - - - - 1.40 0.50 4.20 6.10
Li et al.[20] 128 - - 184.2 - - 2172.0 - - 4.65 4.65
Solano et al.[21] 128 - - 192.0 - - 1831.0 - - 4.21 4.21
Brix et al.[22] 192 - - 196.0 - - 2909.0 - - 6.40 6.40
Lin et al.[23] 256 - - 128.2 - - 985.0 - - 2.06 2.06
Murayama et al.[24] 256 - - - - - - - - 3.50b 3.50b
Dorn et al.[25] 256 - - - - - 723.8 - - 1.67 1.67
Shankar et al.[26] 320 - - - - - 1000.0 - - - -
Manniesing et al.[27] 320 - - - - - - 2.30 2.80 5.00 10.10
Orrison et al.[28] 320 - - - - - - - - 4.30 4.30
Diekmann et al.[29] 320 - - - - - - 1.62 - 3.61 5.23
Salomon et al.[30] 320 - - 400.0b - - 2230.0b - - 5.13b 5.13b
This study 80 55.4 35.2 119.8 912.1 678.6 478.9 1.91 1.42 1.00 4.34
(-): Not mentioned, a Median values, b Minimum values, c Includes two additional exams; CT permeability and post contrast CT
98
(2.07)[12]
(2.22)[15]
(3.70)[17]
(3.90)[16]
(4.90, 4.90)[4,13]
(7.50)[14]
(1.00)[This study]
(1.80)[18]
(4.20,4.21)[21,23]
(4.65)[22]
(6.40)(24)
(1.67)[27]
(2.06)[25]
(3.50)[26] (3.61)[31]
(4.30)[30]
(5.00)[29]
(5.13)[32]
0
1
2
3
4
5
6
7
8
0 64 128 192 256 320 384
ED
(m
Sv)
Slices per rotation
Figure 84. CTP mean ED values for the various
number of slice of the CT systems used
80
[99]
ABBREVIATIONS
ACA: Anterior Cerebral Artery
AEC: Automatic Exposure Control
AIDR: Adaptive Iterative Dose Reduction
BBB: Blood-Brain Barrier
BIC: Bayesian Information Criterion
CBF: Cerebral Blood Flow
CBV: Cerebral Blood Volume
CPP: Comprehensive Prescription Protocol
CT: Computed Tomography
CTA: Computed Tomography Angiography
CTP: Computed Tomography Perfusion
CTDI: Computed Tomography Dose Index
DLP: Dose Length Product
DRL: Diagnostic Reference Level
ED: Effective Dose
FOV: Field Of View
FT: Fourier Transform
FWHM: Full Width at Half Maximum
HU: Hounsfield Units
ICA: internal carotid artery
[100]
ICH: Intracerebral Hemorrhage
ICRP: International Commission on Radiological Protection
MCA: Middle Cerebral Artery
MDCT: Multi Detector Row Computed Tomography
MIP: Maximum-Intensity Projection
MR: Magnetic Resonance
MSAD: Multiple Scan Average Dose
MSCT: Multislice Computed Tomography
MTF: Modulation Transfer Function
MTT: Mean Transit Time
PCA: Posterior Cerebral Artery
PMMA: Polymethyl Methacrylate
PSF: Point Spread Function
ROI: Region of Interest
SAH: Subarachnoid Hemorrhage
SFOV: Scan Field Of View
SVD: Singular Value Decomposition
TLD: Thermo Luminescent Dosimeters
UN-CT: Unenhanced Brain Computed Tomography
[101]
REFERENCES
[1] Wintermark M, Van Melle G, Schnyder P, Revelly J-P, Porchet F, Regli L, et al.
Admission perfusion CT: Prognostic value in patients with severe head trauma.
Radiology 2004;232:211-20.
[2] Wintermark M, Ko NU, Smith WS, Liu S, Higashida RT, Dillon WP. Vasospasm
after subarachnoid hemorrhage: Utility of perfusion CT and CT angiography on
diagnosis and management. Am. J. Neuroradiol. 2006;27:26-34.
[3] Loftus ML, Minkowitz S, Tsiouris AJ, Min RJ, Sanelli PC. Utilization guidelines
for reducing radiation exposure in the evaluation of aneurysmal subarachnoid
hemorrhage: A practice quality improvement project. Am. J. Roentgenol.
2010;195:176-80.
[4] Mnyusiwalla A, Aviv RI, Symons SP. Radiation dose from multidetector row CT
imaging for acute stroke. Neuroradiology 2009;51:635-40.
[5] de Lucas EM, Sánchez E, Gutiérrez A, Mandly AG, Ruiz E, Flórez AF, et al. CT
Protocol for Acute Stroke: Tips and Tricks for General Radiologists. Radiographics
2008;28:1673-87.
[6] Konstas AA, Goldmakher GV, Lee T-Y, Lev MH. Theoretic basis and technical
implementations of CT perfusion in acute ischemic stroke, part 1: Theoretic basis.
Am. J. Neuroradiol. 2009;30:662-8.
[7] Donahue J, Wintermark M. Perfusion CT and acute stroke imaging: Foundations,
applications, and literature review. J. Neuroradiol. 2015;42:21-9.
[8] Othman AE, Afat S, Brockmann MA, Nikoubashman O, Brockmann C, Nikolaou
K, et al. Radiation dose reduction in perfusion CT imaging of the brain: A review of
the literature. J. Neuroradiol. 2016;43:1-5.
[102]
[9] Huang AP-H, Tsai J-C, Kuo L-T, Lee C-W, Lai H-S, Tsai L-K, et al. Clinical
application of perfusion computed tomography in neurosurgery: Clinical article. J.
Neurosurg. 2014;120:473-88.
[10] Sabarudin A, Yusof MZ, Mohamad M, Sun Z. Radiation dose associated with
cerebral CT angiography and CT perfusion: An experimental phantom study. Radiat.
Prot. Dosim. 2014;162:316-21.
[11] Arandjic D, Bonutti F, Biasizzo E, Ciraj-Bjelac O, Floreani M, Giustizieri M, et
al. Radiation doses in cerebral perfusion computed tomography: Patient and phantom
study. Radiat. Prot. Dosim. 2013;154:459-64.
[12] Hoang JK, Wang C, Frush DP, Enterline DS, Samei E, Toncheva G, et al.
Estimation of radiation exposure for brain perfusion CT: Standard protocol
compared with deviations in protocol. Am. J. Roentgenol. 2013;201:W730-W734.
[13] Horiguchi J, Kiura Y, Tanaka J, Fukuda H, Kiguchi M, Fujioka C, et al.
Feasibility of extended-coverage perfusion and dynamic computer tomography (CT)
angiography using toggling-table technique on 64-slice CT. J. Neuroradiol.
2011;38:156-60.
[14] Bricout N, Estrade L, Boustia F, Kalsoum E, Pruvo JP, Leclerc X. Reduced-dose
CT protocol for the assessment of cerebral vasospasm. Neuroradiology
2015;57:1211-8.
[15] Konstas AA, Goldmakher GV, Lee T-Y, Lev MH. Theoretic basis and technical
implementations of CT perfusion in acute ischemic stroke, part 2: Technical
implementations. Am. J. Neuroradiol. 2009;30:885-92.
[16] Fang XK, Ni QQ, Schoepf, UJ, Zhou CS, Chen GZ, Luo S, et al. Image quality,
radiation dose and diagnostic accuracy of 70 kVp whole brain volumetric CT
perfusion imaging: a preliminary study. Eur. Radiol. 2016;26:4184-93.
[103]
[17] International Commission on Radiological Protection. Radiological protection
in medicine. ICRP Publication 105. Ann. ICRP 37, 1-63 (2007).
[18] American Association of Physicists in Medicine. The Measurement, Reporting,
and Management of Radiation Dose in CT. Report 96. AAPM Task Group 23.
American Association of Physicists in Medicine (2008).
[19] Saake M, Goelitz P, Struffert T, Breuer L, Volbers B, Doerfler A, et al.
Comparison of conventional CTA and volume perfusion CTA in evaluation of
cerebral arterial vasculature in acute stroke. Am. J. Neuroradiol. 2012;33: 2068-73.
[20] Li ZL, Li H, Zhang K, Li WJ, Chen X, Wu B, et al. Improvement of image quality
and radiation dose of CT perfusion of the brain by means of low-tube voltage (70 kV).
Eur. Radiol. 2014;24:1906-13.
[21] Corcuera-Solano I, McLellan AM, Doshi AH, Pawha PS, Tanenbaum LN.
Whole-brain adaptive 70-kVp perfusion imaging with variable and extended sampling
improves quality and consistency while reducing dose. Am. J. Neuroradiol.
2014;35:2045-51.
[22] Brix G, Lechel U, Nekolla E, Griebel J, Becker C. Radiation protection issues in
dynamic contrast-enhanced (perfusion) computed tomography. Eur. J. Radiol.
2015;84:2347-58.
[23] Lin CJ, Wu TH, Lin CH, Hung SC, Chiu CF, Liu M-J, et al. Can iterative
reconstruction improve imaging quality for lower radiation CT perfusion? initial
experience. Am. J. Neuroradiol. 2013;34:1516-21.
[24] Murayama K, Katada K, Nakane M, Toyama H, Anno H, Hayakawa M, et al.
Whole-brain perfusion CT performed with a prototype 256-detector row CT system:
Initial experience. Radiology 2009;250:202-11.
[104]
[25] Dorn F, Muenzel D, Meier R, Poppert H, Rummeny EJ, Huber A. Brain
perfusion CT for acute stroke using a 256-slice CT: Improvement of diagnostic
information by large volume coverage. Eur. Radiol. 2011;21:1803-10.
[26] Shankar JJS, Lum C, Sharma M. Whole-brain perfusion imaging with 320-
MDCT scanner: Reducing radiation dose by increasing sampling interval. Am. J.
Roentgenol. 2010;195:1183-6.
[27] Manniesing R, Oei MTH, Van Ginneken B, Prokop M. Quantitative dose
dependency analysis of whole-brain CT perfusion imaging. Radiology 2016;278:190-
7.
[28] Orrison JrWW, Snyder KV, Hopkins LN, Roach CJ, Ringdahl EN, Nazir R, et
al. Whole-brain dynamic CT angiography and perfusion imaging. Clin. Radiol.
2011;66:566-74.
[29] Diekmann S, Siebert E, Juran R, Roll M, Deeg W, Bauknecht H-C, et al. Dose
exposure of patients undergoing comprehensive stroke imaging by multidetector-row
CT: Comparison of 320-detector row and 64-detector row CT scanners. Am. J.
Neuroradiol. 2010;31:1003-9.
[30] Salomon EJ, Barfett J, Willems PWA, Geibprasert S, Bacigaluppi S, Krings T. Dynamic CT Angiography and CT Perfusion Employing a 320-Detector Row CT.
Clin. Neuroradiol. 2009;19:187-96.
[31] International Commission on Radiological Protection. The 2007
Recommendations of the International Commission on Radiological Protection. ICRP
Publication 103. Ann. ICRP 37, 1-35 (2007).
[32] Greek Atomic Energy Commission. 2015 Annual Activity Report (2016).
Available on https://eeae.gr/attachments/article/5774/annual_report_2015_EN.pdf
(accessed on 14 July 2017).
[105]
[33] McNitt-Gray MF. AAPM/RSNA Physics Tutorial for Residents: Topics in CT
Radiation Dose in CT. RadioGraphics 2002;22:1541-53
[34] Nagel HD. CT Parameters that Influence the Radiation Dose. Philips Medical
Systems, Science and Technology, Roentgenstr. 24, D-22335 Hamburg, Germany.
[35] Dance DR, Christofides S, Maidment ADA, McLean ID, Ng KH. Diagnostic
radiology physics. A Handbook for Teachers and Students. International Atomic
Energy Agency (IAEA). Vienna, September 2014;ISBN 978–92–131010–1:1-710