GOUHFI: Generalized and Optimized segmentation tool for Ultra-High Field Images

The Generalized and Optimized segmentation tool for Ultra-High Field Images (GOUHFI) is a deep learning-based fully automatic brain segmentation tool optimized for ultra-high field MRI (i.e., ≥ 7T MRI). Using the domain randomization approach proposed in SynthSeg (Billot et al., 2023), GOUHFI is able to segment images of any contrast, resolution and even field strength, making it broadly applicable across scanners, protocols and imaging centers. 

UPDATE 13/12/2025: GOUHFI 2.0 trained model wieghts are now available! The improved subcortical segmentation tool is available with the `brain_seg.zip` file whereas the cortical parcellation model is the `_parc.zip` one. The original GOUHFI subcortical segmentation model is still available with the `GOUHF.zip` file.