TY - GEN
T1 - STRUDEL
T2 - 12th International Workshop on Machine Learning in Medical Imaging, MLMI 2021, held in conjunction with 24th International Conference on Medical Image Computing and Computer Assisted Intervention, MICCAI 2021
AU - Gröger, Fabian
AU - Rickmann, Anne Marie
AU - Wachinger, Christian
N1 - Publisher Copyright:
© 2021, Springer Nature Switzerland AG.
PY - 2021
Y1 - 2021
N2 - We propose an unsupervised domain adaptation (UDA) approach for white matter hyperintensity (WMH) segmentation, which uses Self-TRaining with Uncertainty DEpendent Label refinement (STRUDEL). Self-training has recently been introduced as a highly effective method for UDA, which is based on self-generated pseudo labels. However, pseudo labels can be very noisy and therefore deteriorate model performance. We propose to predict the uncertainty of pseudo labels and integrate it in the training process with an uncertainty-guided loss function to highlight labels with high certainty. STRUDEL is further improved by incorporating the segmentation output of an existing method in the pseudo label generation that showed high robustness for WMH segmentation. In our experiments, we evaluate STRUDEL with a standard U-Net and a modified network with a higher receptive field. Our results on WMH segmentation across datasets demonstrate the significant improvement of STRUDEL with respect to standard self-training.
AB - We propose an unsupervised domain adaptation (UDA) approach for white matter hyperintensity (WMH) segmentation, which uses Self-TRaining with Uncertainty DEpendent Label refinement (STRUDEL). Self-training has recently been introduced as a highly effective method for UDA, which is based on self-generated pseudo labels. However, pseudo labels can be very noisy and therefore deteriorate model performance. We propose to predict the uncertainty of pseudo labels and integrate it in the training process with an uncertainty-guided loss function to highlight labels with high certainty. STRUDEL is further improved by incorporating the segmentation output of an existing method in the pseudo label generation that showed high robustness for WMH segmentation. In our experiments, we evaluate STRUDEL with a standard U-Net and a modified network with a higher receptive field. Our results on WMH segmentation across datasets demonstrate the significant improvement of STRUDEL with respect to standard self-training.
UR - https://www.scopus.com/pages/publications/85116506222
U2 - 10.1007/978-3-030-87589-3_32
DO - 10.1007/978-3-030-87589-3_32
M3 - Conference contribution
AN - SCOPUS:85116506222
SN - 9783030875886
T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
SP - 306
EP - 316
BT - Machine Learning in Medical Imaging - 12th International Workshop, MLMI 2021, Held in Conjunction with MICCAI 2021, Proceedings
A2 - Lian, Chunfeng
A2 - Cao, Xiaohuan
A2 - Rekik, Islem
A2 - Xu, Xuanang
A2 - Yan, Pingkun
PB - Springer Science and Business Media Deutschland GmbH
Y2 - 27 September 2021 through 27 September 2021
ER -