Better Generalization of White Matter Tract Segmentation to Arbitrary Datasets with Scaled Residual Bootstrap
This work addresses a domain-specific issue in brain connectivity studies by improving segmentation model generalization to arbitrary datasets, though it appears incremental as it builds on existing deep learning methods.
The paper tackles the problem of poor generalization in white matter tract segmentation due to distribution shifts between training and test datasets, and proposes a method using scaled residual bootstrap to augment training data, resulting in consistent improvements across various experimental settings.
White matter (WM) tract segmentation is a crucial step for brain connectivity studies. It is performed on diffusion magnetic resonance imaging (dMRI), and deep neural networks (DNNs) have achieved promising segmentation accuracy. Existing DNN-based methods use an annotated dataset for model training. However, the performance of the trained model on a different test dataset may not be optimal due to distribution shift, and it is desirable to design WM tract segmentation approaches that allow better generalization of the segmentation model to arbitrary test datasets. In this work, we propose a WM tract segmentation approach that improves the generalization with scaled residual bootstrap. The difference between dMRI scans in training and test datasets is most noticeably caused by the different numbers of diffusion gradients and noise levels. Since both of them lead to different signal-to-noise ratios (SNRs) between the training and test data, we propose to augment the training scans by adjusting the noise magnitude and develop an adapted residual bootstrap strategy for the augmentation. To validate the proposed approach, two dMRI datasets were used, and the experimental results show that our method consistently improved the generalization of WM tract segmentation under various settings.