[MINC-users] Classify with the high resolution data

[Andrew Janke]

2009/4/11 vikingy <vikingy at gmail.com>:
> I have some high resolution data (such as 0.6x0.6x0.6) acquired from 7T scanner, and want to perform the segmentation with Classify in minc. It seems that Classify uses the 1x1x1 template. My question is that, would they have to be downsampled to 1x1x1 for use or not? Does the high resolution improve the performance of segmentation?

Hi Bin,

No they will not have to be downsampled. The high res should make the
segmentation "better" but only because there will be less partial
voluming effects with smaller voxels. Classify uses an intensity based
approach so as far as classify is concerned the number of voxesl to
classify is irrelevant.


Andrew Janke