v0.5.0

This updates synthseg to the final commit in the SynthSeg repository.

The main change is in the normalization for parcellated cortical volumes. Previously, these were normalized such that the sum of parcel volumes was the same as the "total cortex" from the main synthseg segmentation. This is now done per-hemisphere, so the sum of left-hemisphere parcels adds up to total_cortex_left and so on for the right hemisphere.

Subcortical segmentations appear to be unafffected in my experiments.

0.4.1

Adds better version provenance and an option to convert synthseg tissue probabilities to ANTs cortical thickness prior classes

0.3.0

Change options to use BIDS terminology.

Full Changelog: v0.2.1...v0.3.0

0.2.1

Optionally resample output back into native space. Synthseg operates on a resampled image that is 1mm isotropic. In reality, it won't resample some images close to 1mm (eg, 0.975x0.975x1mm).

For consistency, we resample everything to exactly 1mm before processing, and then you can resample the output back to the native space.

Full Changelog: v0.1.4...v0.2.1

0.1.4

Runs synthseg with support for using a brain mask to center the input FOV on the brain. Note that synthseg is run on the whole head inside the cropped region, as it is designed to do.