TY - GEN
T1 - 3D Attention U-Net with Pretraining
T2 - 1st Cerebral Aneurysm Detection and Analysis challenge, CADA 2020 held in Conjunction with 23rd International Conference on Medical Image Computing and Computer Assisted Intervention, MICCAI 2020
AU - Su, Ziyu
AU - Jia, Yizhuan
AU - Liao, Weibin
AU - Lv, Yi
AU - Dou, Jiaqi
AU - Sun, Zhongwei
AU - Li, Xuesong
N1 - Publisher Copyright:
© 2021, Springer Nature Switzerland AG.
PY - 2021
Y1 - 2021
N2 - Early detection and accurate segmentation of cerebral aneurysm is important for clinical diagnosis and prevention of rupture, which would be life threatening. 3D images can provide abundant information for characterizing the aneurysm. But the traditional manual segmentation of aneurysms takes lots of time and effort. Therefore, accurate and rapid automatic algorithm for 3D segmentation of aneurysm is needed. U-Net is a widely used deep learning network in medical image segmentation, but its performance is limited by the amount of data. In this challenge of aneurysm segmentation, we proposed to add attention gate and Models Genesis pretraining mechanisms to the classical U-Net model to improve the results. The dice of 3D U-net, 3D Attention U-Net, pretrained 3D U-Net and pretrained 3D Attention U-Net are 0.881, 0.884, 0.890 and 0.907, respectively. The experimental results show that the use of attention gate and Models Genesis can significantly improve the performance of U-Net model in segmenting aneurysms. This work achieved rank one in CADA 2020- Aneurysm Segmentation Challenge.
AB - Early detection and accurate segmentation of cerebral aneurysm is important for clinical diagnosis and prevention of rupture, which would be life threatening. 3D images can provide abundant information for characterizing the aneurysm. But the traditional manual segmentation of aneurysms takes lots of time and effort. Therefore, accurate and rapid automatic algorithm for 3D segmentation of aneurysm is needed. U-Net is a widely used deep learning network in medical image segmentation, but its performance is limited by the amount of data. In this challenge of aneurysm segmentation, we proposed to add attention gate and Models Genesis pretraining mechanisms to the classical U-Net model to improve the results. The dice of 3D U-net, 3D Attention U-Net, pretrained 3D U-Net and pretrained 3D Attention U-Net are 0.881, 0.884, 0.890 and 0.907, respectively. The experimental results show that the use of attention gate and Models Genesis can significantly improve the performance of U-Net model in segmenting aneurysms. This work achieved rank one in CADA 2020- Aneurysm Segmentation Challenge.
KW - 3D Attention U-Net
KW - Image segmentation
KW - Transfer learning
UR - http://www.scopus.com/inward/record.url?scp=85105947613&partnerID=8YFLogxK
U2 - 10.1007/978-3-030-72862-5_6
DO - 10.1007/978-3-030-72862-5_6
M3 - Conference contribution
AN - SCOPUS:85105947613
SN - 9783030728618
T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
SP - 58
EP - 67
BT - Cerebral Aneurysm Detection - First Challenge, CADA 2020, Held in Conjunction with MICCAI 2020, Proceedings
A2 - Hennemuth, Anja
A2 - Goubergrits, Leonid
A2 - Ivantsits, Matthias
A2 - Kuhnigk, Jan-Martin
PB - Springer Science and Business Media Deutschland GmbH
Y2 - 8 October 2020 through 8 October 2020
ER -