TY - JOUR T1 - Evaluation of automated VOI definition freeware tools for PET data analysis of striatal dopaminergic system JF - Journal of Nuclear Medicine JO - J Nucl Med SP - 2045 LP - 2045 VL - 52 IS - supplement 1 AU - Hiroto Kuwabara AU - Pierre-Louis Bazin AU - Ayon Nandi AU - Wichana Chamroonrat AU - Dean Wong Y1 - 2011/05/01 UR - http://jnm.snmjournals.org/content/52/supplement_1/2045.abstract N2 - 2045 Objectives Recent automated volume of interest (VOI) defining tools are expect to facilitate PET data analysis. Four freeware tools, namely TOADS (TS; Bazan et al., 2007), FSL/FIRST (denoted by FF; Patenaude, 2007), freesurfer (FS; Dale et al., 1999), and SPM/unified segmentation (SU; Ashburner and Friston, 2003) were evaluated against a manual method. Methods Twenty pairs of baseline and post-amphetamine 11C-raclopride scans were selected among historical scans of healthy subjects without neuropsychiatric disorders. Putamen (Pu) and caudate nucleus (CN) VOIs were divided into anterior and posterior subdivisions (aPu, pPu, aCN, and pCN) and ventral striatum (vS) (Oswald et al., 2005) when not provided by tools. The VOI overlap (%) was given by the common-to-total volume ratio. Binding potential (BPND) was obtained by MRTM2 (Ichise et al., 2002) using manually defined cerebellum. Amphetamine-induced dopamine release (DArel) was calculated as 1 less post-amphetamine-to-baseline BPND ratio. Results All automated VOIs left room for improvement on visual inspection. Overlaps were excellent for aPu, pPu, and aCN (80-90%) but suboptimal for pCN and vS (50-80%). SU showed greater overlaps than other methods in aPu and pPu, and vS, while TS showed greater overlaps in aCN and pCN (ANOVA followed by t-test; p>0.05). FF, TS, and FS underestimated BPND in all subdivisions (5-10% except pCN), except for FF in vS. SU minimally (<5%) overestimated BPND in aPu and aCN, and underestimated in pCN. All automated methods yielded DArel that were comparable to the manual method: mean estimates differed among methods by less than 2%, except in pCN. Conclusions Automated VOI methods performed differentially among striatum subdivisions, which resulted in over- or under-estimation of regional BPND values compared to the manual method. It appeared questionable to employ automated striatum subdivision VOIs without manual improvements. This study suggested the need for developing fast and accurate correction methods to take full advantage of automated VOI tools ER -