RT Journal Article
SR Electronic
T1 Data Incoherence: A metric to predict optimal detector configuration in PET imaging
JF Journal of Nuclear Medicine
JO J Nucl Med
FD Society of Nuclear Medicine
SP 1849
OP 1849
VO 56
IS supplement 3
A1 Seyyed Majid Valiollahzadeh
A1 John Clark
A1 Osama Mawlawi
YR 2015
UL http://jnm.snmjournals.org/content/56/supplement_3/1849.abstract
AB 1849 Objectives Previously we investigated the image quality with increasing detector gaps while employing compressed sensing reconstruction techniques (SNMMI 2014). Here we investigate the optimal detector configuration for increasing amounts of detector gaps.Methods Detector gaps decrease the number of rows in the PET system matrix (SM). The location of these gaps affect which rows in the system matrix are removed. PET scanners with different detector configuration gaps result in varying system matrices (SM). We propose to use the data incoherence (DIC) of the SM as a metric that can predict the optimal detector configuration in a PET scanner. DIC represents the relationship between acquired samples that correspond to each image pixel index and ranges from zero to one. A DIC close to one implies less correlated samples which denotes a more optimal detector configuration. To test this proposal, we simulated a PET scanner with 26% of its detectors removed according to 4 different configurations: 1) equidistant detector gaps along the circumference; 2) three detector gaps at 0, 120, and 240°; 3) two detector gaps at 0 and 180°; and 4) two detector gaps at 90 and 270°. The DIC for all voxel values was then calculated, plotted, and compared for the four sampling patterns. This process was then repeated for the same PET scanner but with 40% and 62% detectors removed. We also validated this approach using phantom and patient studies.Results The table shows the mean DIC values for the different amounts of detector removal and corresponding sampling patterns. Visual inspection also showed that patterns 3 and 4 had the highest and most uniform DIC distribution suggesting optimal configuration. Pattern 2 had the worst performance. Results from phantom and patient studies will be presented.Conclusions These results indicate that DIC might be able to predict the optimal detector gap configuration in a PET scanner. DIC of different patterns and percent detector removal