TrackletGPT: A Language-like GPT Framework for White Matter Tract Segmentation
This work addresses the complex problem of white matter tract segmentation for brain connectivity studies and neurological applications, representing an incremental improvement with a novel method for a known bottleneck.
The authors tackled white matter tract segmentation by proposing TrackletGPT, a language-like GPT framework that uses tracklets to reintroduce sequential information, achieving superior performance on DICE, Overlap, and Overreach scores compared to state-of-the-art methods on TractoInferno and HCP datasets.
White Matter Tract Segmentation is imperative for studying brain structural connectivity, neurological disorders and neurosurgery. This task remains complex, as tracts differ among themselves, across subjects and conditions, yet have similar 3D structure across hemispheres and subjects. To address these challenges, we propose TrackletGPT, a language-like GPT framework which reintroduces sequential information in tokens using tracklets. TrackletGPT generalises seamlessly across datasets, is fully automatic, and encodes granular sub-streamline segments, Tracklets, scaling and refining GPT models in Tractography Segmentation. Based on our experiments, TrackletGPT outperforms state-of-the-art methods on average DICE, Overlap and Overreach scores on TractoInferno and HCP datasets, even on inter-dataset experiments.