Cerebral blood flow in depressed patients: a methodological comparison of statistical parametric mapping and region of interest analyses.

Functional brain imaging has assumed a leading role in neuropsychiatric research. However, findings reported for mental disorders often vary. Whether this reflects diversity in pathophysiology or heterogeneity of imaging techniques and data-analytic procedures is still unknown. This study compares region of interest (ROI) and statistical parametric mapping (SPM) analyses of a Tc99m-HMPAO single photon emission computed tomography (SPECT) imaging study of 23 depressed and 21 control subjects. Reduced regional cerebral blood flow (rCBF) was demonstrated by both methods in the right parietal and occipital lobes, but additional regions were identified only on ROI analysis (left temporal) and only on SPM analysis (left parietal). To investigate the contribution of SPM spatial normalization to these discrepancies, further ROI analyses were performed, applying the original ROI templates to normalized images, and applying regions identified by SPM to the original images. This study demonstrated considerable overlap in findings of SPM and ROI analyses. Differences between these methods may be mostly related to subjective placement of ROIs in ROI analysis, and standardized warping inherent in normalization in SPM. Given the advantages and drawbacks of each procedure, the choice of methodology should be determined in accordance with the study design, and complementary use of both methods may be considered. Copyright 2002 Elsevier Science Ireland Ltd.