Label merge-and-split: A graph-colouring approach for memory-efficient brain parcellation
This addresses memory and efficiency issues in deep learning for brain parcellation, offering a practical solution for medical imaging tasks with many spatially separate classes, though it is incremental as it builds on existing atlas-based priors.
The paper tackles the challenge of whole brain parcellation with many labels by introducing a label merge-and-split method that reduces the number of labels by up to 68% using graph coloring, achieving comparable segmentation accuracy to baseline methods while significantly cutting training time, inference time, and GPU memory usage.
Whole brain parcellation requires inferring hundreds of segmentation labels in large image volumes and thus presents significant practical challenges for deep learning approaches. We introduce label merge-and-split, a method that first greatly reduces the effective number of labels required for learning-based whole brain parcellation and then recovers original labels. Using a greedy graph colouring algorithm, our method automatically groups and merges multiple spatially separate labels prior to model training and inference. The merged labels may be semantically unrelated. A deep learning model is trained to predict merged labels. At inference time, original labels are restored using atlas-based influence regions. In our experiments, the proposed approach reduces the number of labels by up to 68% while achieving segmentation accuracy comparable to the baseline method without label merging and splitting. Moreover, model training and inference times as well as GPU memory requirements were reduced significantly. The proposed method can be applied to all semantic segmentation tasks with a large number of spatially separate classes within an atlas-based prior.