[SPM analysis on PET and SPECT data].

I described the methodological explanation of statistical parametric mapping (SPM) with the special attention to the clinical applications. SPM was designed mainly by K. Friston and R. Frackowiack to visualize the statistically significant regions on the CBF data sets obtained from PET or SPECT during a various kinds of activation. In short, images were realigned to the first scan to eliminate the head position movements, and were normalized to the standard brain shape (Talairach & Tournoux) by linear or non-linear transformation. After that, gausian filtering was done in order to eliminate the individual brain differences and to increase the signal to noise ratio. The pixel by pixel analysis was done with multiple comparison correction, and the statistically significant areas were displayed. In clinical applications, comparison of normal controls with Alzheimer's disease, diffuse Lewy body disease, and some activation study in the pathological vs normal condition. In SPECT, we can obtain the brain perfusion images by the injection of Tc-99m labeled CBF tracers, so that we can analyze the condition with the subjects not confined to the scanner.