Mappa funzionale della corteccia motoria e sensoriale primaria con tecniche di Risonanza Magnetica. II. Tecniche di analisi di immagine

Functional Magnetic Resonance Imaging (fMRI) techniques to investigate brain function are now available on clinical MR systems. However, the software packages provided with the MR equipment to analyze the functional images are often inadequate. In the present study, two registration algorithms for correcting motion artifacts and three procedures of statistical analysis (t-test, correlation analysis, Kolmogorov-Smirnov test) were compared using programs implemented on a graphic workstation. For both registration algorithms, transformation parameters for in plane translations and rotation of images were significantly affected by the task, being higher during sequential finger movements than during the control (visual imagery) condition. Regions of interest were identificated on the anatomical images and their boundaries automatically projected on functional images. The number of significantly activated pixels in the pre- and postcentral areas was not significantly different after the registration with the two procedures. The percentage of pixels of the pre- and postcentral areas whose signal intensity was significantly different between the two tasks decreased with respect to the adopted threshold of significance as a power function. For an area identified outside the brain, the same relation was linear; no activated pixel was found for p < 0.001. The application of the t-test or of the correlation analysis yielded similar results. The analysis of the profile of mean normalized signal intensity showed higher increases in signal intensities during the motor task in the precentral gyrus than in the postcentral gyrus. This appears to be due to a greater number of activated pixels during motor performance. The application of registration procedures, the identification of the regions of interest on the basis of the anatomical images and appropriate statistical analyses allow a more detailed characterization of task-related activation.