The application of statistical parametric mapping to 123I-FP-CIT SPECT in dementia with Lewy bodies, Alzheimer's disease and Parkinson's disease

Dopaminergic loss can be visualised using I-123-FP-CIT single photon emission computed tomography (SPECT) in several disorders including Parkinson's disease (PD) and dementia with Lewy bodies (DLB). Most previous SPECT studies have adopted region of interest (ROI) methods for analysis, which are subjective and operator-dependent. The purpose of this study was to investigate differences in striatal binding of I-123-FP-CIT SPECT using the automated technique of statistical parametric mapping (SPM99) in subjects with DLB, Alzheimer's disease (AD), PD and healthy age-matched controls. This involved spatial normalisation of each subject's image to a customised template, followed by smoothing and intensity normalisation of each image to its corresponding mean occipital count per voxel. Group differences were assessed using a two-sample t test. Applying a height threshold of P less than or equal to 0.05 corrected, the SPM(t) maps showed a significant bilateral reduced uptake in caudate, anterior and posterior putamen in DLB and PD subjects compared to AD subjects and controls. Significant reduction in binding was also observed bilaterally in the caudate nucleus in AD compared to controls. Striatal binding was indistinguishable between patients with DLB and PD. To investigate the usefulness of SPM as a decision aid in the evaluation of visually rated normal and abnormal patterns of uptake, receiver operator characteristic (ROC) curve analysis was performed using data from single-subject SPMs. The areas under the ROC curves were greater than 0.92, demonstrating comparable discriminatory power with visual rating. The automated voxel-based approach is a viable alternative to the subjective and often time-consuming method of ROI and, in addition, may have the potential to differentiate between normal and abnormal patterns of uptake in a manner similar to visual inspection. (C) 2004 Elsevier Inc. All rights reserved.