Assessment of somatostatin image quantification with SPET and SPET-CT to aid characterisation of disease.
PhD thesis, University of Glasgow.
Full text available as:
Work was undertaken in this thesis to assess the use of somatostatin image quantification with SPET and SPET-CT to aid the characterisation of disease within the body. Two radionuclide somatostatin analogues were used for this assessment, the first was NeoSPECT and the second, OctreoScan.
The primary aim of work in this thesis was to assess the role of NeoSPECT imaging in the characterisation of disease within the lungs, that is, to differentiate benign from malignant disease. Two forms of image quantification were used in the NeoSPECT assessment, a tumour to background ratio (T:B) and a value of tumour percentage uptake (% uptake).
Values of T:B and % uptake were calculated from SPET images acquired 2 hours post injection. T:B results from the benign group (n = 8) demonstrated a median T:B of 2.21, whilst the malignant group (n = 28) demonstrated a median T:B of 2.01. The differences between the groups were tested statistically via a Mann-Whitney test, which showed there to be no statistical difference between the groups (p=0.90, 95.4% CI of (-0.5598, 0.5498)).
To undertake the calculation of % uptake a non-patient acquisition (a standard acquisition) was also required, unfortunately not all of the patient cohort used for the T:B assessment had this additional acquisition. As a result of this numbers were low for the % uptake assessment in patients with a benign histology (n = 2), therefore statistical analysis could not be performed. However, review of the range of values for each histology within the malignant group proved useful as no differences were demonstrated between the ranges of values which could help to differentiate between the histologies.
Quantification of dual time point imaging was also assessed to determine if there were any variations in values calculated that could also help differentiate the different histologies. For this assessment patients were images at 2 and 4 hours post injection. Results from the Wilcoxon Signed Rank test of the T:B assessment found there to be no statistically significant difference between values of T:B calculated at 2 and 4 hours that was characteristic of tumour type (p=1.0 and p=0.14). The difference in % uptake between 2 and 4 hours was also assessed via a Wilcoxon Signed Rank test, this test also concluded there to be no significant difference value of % uptake between the two acquisitions of the malignant group (p = 0.73).
An attempt was also made to quantify ‘other’ uptake within the mediastinum, however, a lack of anatomical information made correlation with histology impossible and as a result no firm conclusions relating image quantification to histology could be drawn from this work.
Work from this thesis concluded no quantitative difference between tissue histology could be demonstrated using NeoSPECT, either from single or dual time point imaging.
As a result of the NeoSPECT work a number of factors which limited the accuracy and reproducibility of SPET image quantification were identified. Towards the end of the NeoSPECT work hybrid imaging (SPET-CT) became available within the department at Glasgow Royal Infirmary. It was believed that hybrid imaging could resolve some of the limitations and subsequently improve the accuracy of SPET image quantification. However, NeoSPECT was removed from the market for a short period of time and therefore a similar somatostatin analogue, OctreoScan, was used to investigate if the accuracy of somatostatin image quantification could be improved as a result of SPET-CT and its associated reconstruction algorithms including a CT based attenuation correction.
Firstly, a qualitative assessment of image quality using the new hybrid reconstructions techniques was undertaken via an observer study. Images were reconstructed with the existing reconstruction techniques, as used for the NeoSPECT work, and with the new hybrid imaging techniques. Four experienced observers blinded to reconstruction technique were asked to score images in terms of their overall image quality. A Friedman test was performed on the scores for each observer, three of the four observers demonstrated a statistically significant difference in their scores between the existing and new hybrid technique (p = 0.00, p = 0.003, p= 0.00), with the new hybrid technique being assigned the highest scores in terms of image quality.
Images were also assessed semi-quantitatively via profile analysis which also demonstrated a clear differentiation between the existing and new hybrid techniques with increased image quality being demonstrated in the hybrid data set.
The quantitative accuracy of hybrid imaging was also assessed using phantom data. For 111In the difference of the value of absolute activity calculated and that measured varied by 35% but this improved to 21 % when scatter and CT-attenuation based corrections were applied. For 99mTc a much more notable difference between the existing techniques used in chapter 2 and those available from the use of hybrid imaging was demonstrated, the difference in the value of absolute activity calculated and that measured improved from 67% to 0.04%, respectively.
Work in this thesis clearly demonstrated an improvement in image quality and accuracy in SPET quantification as a result of hybrid imaging techniques.
Actions (login required)