TY - GEN
T1 - Improved White Matter Tract Segmentation for Unannotated Dataset
AU - Zeng, Yijia
AU - Liu, Wan
AU - Liu, Zhiwen
AU - Ye, Chuyang
N1 - Publisher Copyright:
© 2022 ACM.
PY - 2022/10/21
Y1 - 2022/10/21
N2 - White matter (WM) tract segmentation is beneficial to brain research, which provides a valuable tool for analyzing brain development and disease. The introduction of convolutional neural networks (CNNs) has greatly improved the accuracy of WM tract segmentation. However, the training of CNNs usually requires extensive manual annotations of WM tracts, which are often difficult to obtain in practical applications. Therefore, in this study, we explore two methods to realize CNN-based WM tract segmentation when there are no manual annotations of WM tracts for the target dataset and improve the segmentation accuracy. The first method generates registration-based pseudo labels for the target dataset to train the WM tract segmentation network. Specifically, we register images of the publicly available annotated dataset to images of the unlabeled target dataset and improve the binarization strategy by taking advantage of the characteristics of registration and WM tracts to generate the soft labels of WM tracts for target dataset. Moreover, we propose the other method to construct loss weighted matrix for network training using the registration information, which reduces the impact of registration error and further improves the segmentation performance. We evaluated the proposed methods with two dMRI datasets. The results show that the proposed methods are effective in improving the segmentation performance of WM tracts when the manual annotations are unavailable.
AB - White matter (WM) tract segmentation is beneficial to brain research, which provides a valuable tool for analyzing brain development and disease. The introduction of convolutional neural networks (CNNs) has greatly improved the accuracy of WM tract segmentation. However, the training of CNNs usually requires extensive manual annotations of WM tracts, which are often difficult to obtain in practical applications. Therefore, in this study, we explore two methods to realize CNN-based WM tract segmentation when there are no manual annotations of WM tracts for the target dataset and improve the segmentation accuracy. The first method generates registration-based pseudo labels for the target dataset to train the WM tract segmentation network. Specifically, we register images of the publicly available annotated dataset to images of the unlabeled target dataset and improve the binarization strategy by taking advantage of the characteristics of registration and WM tracts to generate the soft labels of WM tracts for target dataset. Moreover, we propose the other method to construct loss weighted matrix for network training using the registration information, which reduces the impact of registration error and further improves the segmentation performance. We evaluated the proposed methods with two dMRI datasets. The results show that the proposed methods are effective in improving the segmentation performance of WM tracts when the manual annotations are unavailable.
KW - Pseudo labels
KW - Registration
KW - Weighted matrix
KW - White matter tract segmentation
UR - http://www.scopus.com/inward/record.url?scp=85145565949&partnerID=8YFLogxK
U2 - 10.1145/3569966.3570076
DO - 10.1145/3569966.3570076
M3 - Conference contribution
AN - SCOPUS:85145565949
T3 - ACM International Conference Proceeding Series
SP - 388
EP - 392
BT - CSSE 2022 - 2022 5th International Conference on Computer Science and Software Engineering
PB - Association for Computing Machinery
T2 - 5th International Conference on Computer Science and Software Engineering, CSSE 2022
Y2 - 21 October 2022 through 23 October 2022
ER -