Quantitative Accuracy of Iterative Reconstruction Algorithms in Positron Emission Tomography A ...

192 Pages·2017·14.59 MB·English

Quantitative Accuracy
of Iterative Reconstruction Algorithms
in Positron Emission Tomography

A thesis submitted to the
University of Manchester
for the degree of Doctor of Philosophy in
the Faculty of
Biology, Medicine and Health

2016

Ian Armstrong

School of Health Sciences
Contents
List of figures...............................................................................................................4
List of tables.................................................................................................................6
List of abbreviations.....................................................................................................7
Abstract........................................................................................................................9
Declaration.................................................................................................................10
Copyright statement...................................................................................................11
Acknowledgements....................................................................................................12
The Author.................................................................................................................13
1. Introduction........................................................................................................14
1.1 Cancer........................................................................................................14
1.2 The birth of Positron Emission Tomography.............................................16
1.3 Quantification of uptake of radio-labelled compounds in tumours...........16
1.4 The growth of PET and PET/CT imaging and its applications..................20
1.5 PET/CT in the UK and NHS......................................................................26
1.6 PET in Manchester.....................................................................................28
1.7 Scope of this project...................................................................................30
1.8 Thesis structure..........................................................................................32
1.9 References..................................................................................................42
2. Principles of Positron Emission Tomography................................................50
2.1 Overview....................................................................................................50
2.2 Gamma-ray interactions.............................................................................51
2.3 PET detector design...................................................................................52
2.4 Development of PET hardware..................................................................55
2.5 Factors that degrade PET data....................................................................57
2.6 Advances in PET design............................................................................62
2.7 Time of flight PET.....................................................................................64
2.8 References..................................................................................................67
3. PET image acquisition and reconstruction.........................................................72
3.1 Overview....................................................................................................72
3.2 PET data collection....................................................................................73
3.3 Data corrections.........................................................................................77
3.4 Image reconstruction..................................................................................84
3.5 Analytical reconstruction...........................................................................84
2
3.6 Iterative reconstruction...............................................................................87
3.7 References................................................................................................100
4. First paper............................................................................................................109
5. Second paper........................................................................................................138
6. Third paper...........................................................................................................157
7. Fourth paper.........................................................................................................168
8. Summary and conclusions................................................................................177
8.1 Impact on quantification..........................................................................177
8.2 Improvements in image quality................................................................180
8.3 Conclusions..............................................................................................182
8.4 References................................................................................................184
9. Future work after PhD......................................................................................185
9.1 Correlation with lung nodule and lymph node histology.........................185
9.2 Continuation of development of extended NEMA phantom...................187
9.3 Assessment of image quality with next-generation PET scanners...........189
9.4 References................................................................................................191

Thesis word count: 25,790 (excluding papers and references)
3
LIST OF FIGURES
Figures in Chapter 1: Introduction
Figure 1.1 Illustration of activity concentration of FDG in blood and 19
tissue over time
Figure 1.2 NEMA image quality phantom 34
Figures in Chapter 2: Principles of Positron Emission Tomography
Figure 2.1 Block detector design 56
Figure 2.2 Different coincidence event types 60
Figure 2.3 Illustration of the statistical noise observed in images 61
Figure 2.4 Representation of depth of interaction and inter-crystal 62
scatter
Figures in Chapter 3: PET image acquisition and reconstruction
Figure 3.1 An example PET image 72
Figure 3.2 Example of a sinogram 74
Figure 3.3 Example of direct-plane and cross-plane sinograms 75
Figure 3.4 Illustration of 2D and 3D data acquisition 76
Figure 3.5 Assignment of events into TOF sinograms 77
Figure 3.6 Distribution of scattering angle cross-section for different 80
gamma-ray energies
Figure 3.7 Example of backprojection for a varying range of 85
projection angles
Figure 3.8 Schematic representation of the ramp filter 86
Figure 3.9 Illustration of filtered backprojection 86
Figure 3.10 Representation of depth of interaction and inter-crystal 94
scatter and resultant image
Figure 3.11 Schematic illustration of the Gibbs artefact 96
Figures in Chapter 4: First paper
Figure 1 Transaxial PET images of the NEMA IQ phantom 116
Figure 2 Max-AC recovery for each of the six spheres filled in the 118
8:1 phantom
Figure 3 Max-AC values where the reconstruction parameters are 120
set to minimize the variation of recovery across the
different sphere sizes
Figure 4 AC recovery for the 10 mm and 37 mm sphere for max- 122
AC, mean-AC and peak-AC
Figure 5 Coefficient of variation of AC measurements for the 124
10 mm and 37 mm sphere for max-AC, mean-AC and
peak-AC
Figure 6 Max-AC for the 10 mm and 37 mm sphere filled with 126
2:1, 4:1 and 8:1 (bottom row) concentration ratio
Figure 7 Activity concentration recovery for each sphere after 3 129
iterations of each reconstruction algorithm
4
Figures in Chapter 5: Second paper
Figure 1 Phantom sphere profiles 146
Figure 2 Coronal PET images 146
Figure 3 Bland-Altman plots of relative differences for SUV 149
max
relative to OSEM
Figure 4 Bland-Altman plots of relative differences for SUV 149
peak
relative to OSEM
Figure 5 Bland-Altman plots of relative differences for TLG-40 150
relative to OSEM
Figure 6 Percentage differences versus lesion volume 150
Figures in Chapter 6: Third paper
Figure 1 Coronal and transaxial PET images for patient with BMI 162
of 26.6 kg/m2
Figure 2 Coronal and transaxial PET images for patient with BMI 163
of 42.4 kg/m2
Figure 3 Plots of measured SNR in the liver 164
Figure 4 Categorical visual assessments of image quality 165
Figure 5 Noise-equivalent count rate against 18F-FDG activity in 166
the NEMA phantom
Figures in Chapter 7: Fourth paper
Figure 1 Graphical representation of the estimation of ensemble 170
variance
Figure 2 Sections through various 3-D regions that were evaluated 170
Figure 3 Coefficient of variation of EV values of spherical regions 171
Figure 4 Estimated covariance kernels from 200 replicate images 172
Figure 5 Comparison of estimated and measured EV for spherical 172
regions
Figure 6 Repeatability of EV measures using data from separate 173
phantoms
Figure 7 Spatial variation of covariance kernels 174
Figure 8 EV values calculated from different locations in the 173
phantom
Figure 9 Comparison of estimated EV from NEMA background 173
Figure 10 Ratios of measured EV in background and NEMA 175
spheres with various reconstruction parameters
Figure 11 Covariance kernels derived from four patient livers 175
Figure 12 Comparison of estimated EV values using covariance 174
kernels from four patient livers
Figures in Chapter 8: Summary and conclusions
Figure 8.1 Gains in SNR using TOF for 88 routine patients 181
5
LIST OF TABLES
Tables in Chapter 1: Introduction
Table 1.1 Top ten causes of registered male deaths in England and 14
Wales for 2014
Table 1.2 Top ten causes of registered female deaths in England 15
and Wales for 2014
Table 1.3 T, N and M classifications based on disease 23
characterisation for lung cancer
Table 1.4 Disease staging of lung cancer according to T, N and M 24
classifications
Tables in Chapter 2: Principles of Positron Emission Tomography
Table 2.1 Characteristics of positron-emitting radionuclides that are 51
commonly used in PET imaging
Table 2.2 Physical characteristics of scintillation crystals for PET 55
detectors
Tables in Chapter 4: First paper
Table 1 Reconstruction parameters to give minimum dependency 119
of max-AC on sphere size
Tables in Chapter 5: Second paper
Table 1 Phantom recovery data in unfiltered images 145
Table 2 Phantom recovery data 145
Table 3 Patient liver noise 147
Table 4 Relative uptake differences for matched voxel COV 148
Table 5 Relative uptake differences for matched SUV 148
max
Table 6 SUV changes for lesions with borderline values for 151
max
malignancy
Tables in Chapter 6: Third paper
Table 1 Patient demographics in study group 160
Table 2 Liver SNR, mediastinum SUV and lesion SUV for 163
mean max
image data
Table 3 Relative changes of liver SNR, mediastinum SUV 164
mean
and lesion SUV for image data
max
Table 4 Summary of reductions in prescribed activity and 166
scanning time
6
LIST OF ABBREVIATIONS
AC Activity concentration
ATP Adenosine triphosphate
BGO Bismuth germinate
CMUH Central Manchester University Hospitals
COV Coefficient of variation
CRG Clinical Reference Group
CT Computed tomography
DG Deoxyglucose
DG-6-P Deoxyglucose-6-phosphate
EV Ensemble variance
FBP Filtered back-projection
FDEV Fourier-derived ensemble variance
FDG Fluorodeoxyglucose
FOV Field of view
FWHM Full-width half-maximum
GE General Electric
HD HD·PET (Siemens reconstruction algorithm)
ICSCNM Intercollegiate Standing Committee on Nuclear Medicine
IQ Image quality (PET phantom)
LAC Linear attenuation coefficient
LOR Line of response
LSO Lutetium oxyorthosilicate
MLEM Maximum likelihood expectation maximisation
MR Magnetic resonance
MRC Medical Research Council
NEC Noise equivalent counts
NEMA National Electrical Manufacturers Association
NHS National Health Service
NICE National Institute for Clinical Excellence
OP Ordinary Poisson
OSEM Ordered subset expectation maximisation
PERCIST PET Response Criteria in Solid Tumours
PET Positron emission tomography
PET/CT PET with integrated CT
PET/MR PET with integrated MR
PMT Photomultiplier tube
PSF Point spread function
RM Resolution modelling
ROI Region of interest
SNR Signal-to-noise ratio
7
SUV Standardised uptake value
TLG Total lesion glycolysis
TOF Time of flight
TNM Tumour, Node, Metastases (cancer scoring system)
UHD ultraHD·PET (Siemens reconstruction algorithm)
VOI Volume of interest
WMIC Wolfson Molecular Imaging Centre
8
ABSTRACT
The University of Manchester
Ian Armstrong. Doctor of Philosophy.
Quantitative Accuracy of Iterative Reconstruction Algorithms in Positron Emission
Tomography
2016

Positron Emission Tomography (PET) plays an essential role in the management of
patients with cancer. It is used to detect and characterise malignancy as well as
monitor response to therapy. PET is a quantitative imaging tool, producing images
that quantify the uptake of a radiotracer that has been administered to the patient. The
most common measure of uptake derived from the image is known as a Standardised
Uptake Value (SUV). Data acquired on the scanner is processed to produce images
that are reported by clinicians. This task is known as image reconstruction and uses
computational algorithms to process the scan data. The last decade has seen
substantial development of these algorithms, which have become commercially
available: modelling of the scanner spatial resolution (resolution modelling) and time
of flight (TOF). The Biograph mCT was the first scanner from Siemens Healthcare
to feature these two algorithms and the scanner at Central Manchester University
Hospitals was the first Biograph mCT to go live in the UK. This PhD project,
sponsored by Siemens Healthcare, aims to evaluate the effect of these algorithms on
SUV in routine oncology imaging through a combination of phantom and patient
studies.
Resolution modelling improved visualisation of small objects and resulted in
significant increases of uptake measurements. This may pose a challenge to
clinicians when interpreting established uptake metrics that are used as an indication
of disease status. Resolution modelling reduced the variability of SUV. This
improved precision is particularly beneficial when assessing SUV changes during
therapy monitoring.
TOF was shown to reduce image noise with a conservation of FDG uptake
measurements, relative to non-TOF algorithms. As a result of this work, TOF has
been used routinely since mid-2014 at the CMUH department. This has facilitated a
reduction of patient and staff radiation dose and an increase of 100 scans performed
each year in the department.
9
DECLARATION
No portion of the work referred to in the thesis has been submitted in support of an
application for another degree or qualification of this or any other university or other
institute of learning.
10