Hi Zhengchen

I don't think there is a general answer to your question as it depends on how noisy the input image is, what denoising algorithm you use and what outcome measures you care about. For example, it might improve putamen/pallidum but worsen thickness measures in low contrast regions such as motor cortex.

cheers Bruce On Tue, 27 Sep 2016, Zhengchen Cai wrote:

Hi Freesurfers, In order to get the better segmentation and surfaces by Freesurfer, do you think it is necessary to denoise T1 before feed it to Freesurfer? Since there are several advanced MRI denoising methods which are not covered in FS preprocessing, I though it would be better to perform one of them before FS. However, one of my concern is that those denoising will modify the intensity of the image and this may effect or even conflict with the processing in FS such as intensity normalization or registration to the template. Based on my experiences of FSL, most of time, the denoising methods will make the segmentation worse. For FS, I've tried just few subjects and the segmentations were better than feeding the raw MRIs. What is your suggestion in general case? or maybe it is not reasonable at all to denoise before FS? Many thanks.

Best regars, Zhengchen