Dear Jose,
The optimal choice will also depend on whether you have additional data (e.g. T2, FLAIR), whether you want to label the lesions manually or not, whether you just want to segment automatically or also normalise into MNI space (as possibly, it's not just a patient but an old patient, with the preprocessing maybe benefiting from age-appropriate templates/tissue probability maps), which type of lesion it is, ... In any case, there are several toolboxes dealing with lesion data which might be better than the default SPM routines, e.g. LST http://www.applied-statistics.de/lst.html , Multiple Sclerosis lesion segmentation toolbox http://atc.udg.edu/salem/slsToolbox/index.html (might work with other lesions as well), the Clinical toolbox http://www.nitrc.org/plugins/mwiki/index.php/clinicaltbx:MainPage
If you rather aim at working with the "best" methods available it might be worth to have a look at latest algorithms and their outcomes from recent challenges like the Ischemic Stroke Lesion Segmentation Challenge, Multimodal Brain Image Tumor Segmentation, Longitudinal Multiple Sclerosis Lesion Segmentation Challenge.
Best
Helmut
|