Dear FSL expert, I have a question about the use of MELODIC to analysis the SPECT data. Our lab try to use the ICA method to analysis the SPECT image. We have 43 control subjects and 7 patients. we put these image files into a 4d analyze image file and the number of control subject are 1-43, patients are 44-50. the threshold is setting in 15% and the output component is set 12. My question is how to choice a component which may be different between control and subjects ? can I use the group test (t-test) to compare the data in raw 1-43 and 44-47 in each colomn from melodic_mix. Thank your opinion. Sincerely Dr. Jung Lung hsu Department of Neurology Taiwan