A comparison between the conventional manual ROI method and an automatic algorithm for semiquantitative analysis of SPECT studies

In this study, the performance of a free software for automatic segmentation of striatal SPECT brain studies (BasGanV2 – www.aimn.it) and a standard manual Region Of Interest (ROI) method were compared. The anthropomorphic Alderson RSD phantom, filled with solutions at different concentration of 123I-FP-CIT with Caudate-Putamen to Background ratios between 1 and 8.7 and Caudate to Putamen ratios between 1 and 2, was imaged on a Philips-Irix triple head gamma camera. Images were reconstructed using filtered back-projection and processed with both BasGanV2, that provides normalized striatal uptake values on volumetric anatomical ROIs, and a manual method, based on average counts per voxel in ROIs drawn in a three-slice section. Caudate-Putamen/Background and Caudate/Putamen ratios obtained with the two methods were compared with true experimental ratios. Good correlation was found for each method; BasGanV2, however, has higher R index (BasGan Rmean = 0.95, p < 0.001) than the ROI method (Rmean = 0.89, p < 0.005). The results obtained suggest that this new software can be suitable in daily practice, being an accurate, simple, fully automatic and reproducible method of semiquantitative analysis of 123I-FP-CIT SPECT data with, moreover, the advantage of the availability of a control subject's database.