Reply-To: | | [log in to unmask][log in to unmask]> wrote:
> Hi all, > In experimenting with SPM12 segmentations, we have found some areas of the > brain that are often misclassified as GM, despite the TPM's correctly > showing zero possibility of GM in these voxels. This suggests that things > could potentially improve by tuning the algorithm's balance between > trusting the Gaussian image intensity likelihoods vs the TPM's acting as > priors. After examining the Ashburner and Friston Neuroimage paper on > Unified Segmentation, spm12 toolbox help text, and the available parameters > in the toolbox interface, we suspect that this may not be readily > alterable. Perhaps we are missing something. Does anyone know a way that we > could experiment with adjusting such a balance? > Thank you very much, > Christopher Schwarz > > -- > Christopher G. Schwarz, Ph.D. | Research Associate, Diagnostic Radiology | > Opus Center for Advanced Imaging Research | 507-538-4867 | > [log in to unmask] > Mayo Clinic | 200 First Street SW | Rochester, MN 55905 | > http://mayoclinic.org >[log in to unmask] |