Page 90 - Quantitative Imaging of Small Tumours with Positron Emission Tomography
P. 90

                                 Image analysis All suspicious FDHT-avid lesions with uptake exceeding background were included. Volumes of interest (VOI) were delineated on the original PET images using a semi-automatic algorithm using a threshold of 50% of the peak value within the area of interest with correction for local background uptake(23). From each VOI, we derived the average, peak, and maximum activity concentrations (AC; Bq/cc). Next, SUV was derived by dividing the tumour AC to a normalization factor. Two normalizations were used (12): a) injected dose per kg bodyweight (bw), and b) area-under-the-curve of the parent plasma calibrated image-derived input function (AUC-PP). The AR-positive tumour volume (ARTV; mL) was 4 defined as the sum of all voxel volumes within a VOI. To assess lesion detectability, we generated a single voxel thick shell around the tumour VOI to determine local background activity, yielding the contrast-to-noise ratio (CNR) as follows: Eq. 2 where ACavg is the average tumour AC; ACbgr is the average background AC, and SDbgr is the standard deviation of AC in voxels included in the background shell. To compare image noise levels of the 100% and 50% count images, a 3cm diameter spherical VOI was placed in the liver, from which the coefficient of variation (COV%) was calculated as follows: Eq. 3 where SDliver is the standard deviation of the ACs of voxels within the liver VOI, and ACliver is the mean AC of voxels within the liver VOI. Statistical Analysis Analyses were performed using SPSS statistics (v22, IBM) and Excel datasheets. Intrascan variability was defined as the difference in SUVs between the split scans of each original scan (eg. split 1 versus split 2). Interscan variability was defined as the test-retest variability (repeatability) of SUVs. Both intra- and interscan variability were assessed on a per-lesion basis. Repeatability coefficients (RC%) were calculated from the standard deviations of the relative differences of measured SUVs between test-retest scans (day 1 vs day 2) and between split [18F]FDHT PET repeatability                                                89       


































































































   88   89   90   91   92