Loss odyssey in medical image segmentation
Loss odyssey in medical image segmentation
复制标题
DOI：
10.1016/j.media.2021.102035
复制
发表时间：
2021-04-01
影响因子：
10.9
通讯作者：
Martel, Anne L.
中科院分区：
文献类型：
--
作者：
Ma, Jun;Chen, Jianan;Martel, Anne L.
损失函数是基于深度学习的分割方法的重要组成部分。在过去的五年中，已经提出了许多损失函数用于各种分割任务。然而，缺乏对这些损失函数的效用的系统研究。在本文中，我们提出了一个全面的审查分割损失函数的组织方式。我们还对四个典型的3D分割任务的20个一般损失函数进行了首次大规模分析，涉及来自10多个医疗中心的六个公共数据集。结果表明，没有一个损失可以在四个分割任务上始终实现最佳性能，但复合损失函数（例如，具有TopK损失的骰子，焦点损失，Hausdorff距离损失和边界损失）是最鲁棒的损失。我们的代码和分割结果是公开的，可以作为损失函数基准。我们希望这项工作也将为社区提供新的损失函数开发的见解。(c)2021爱思唯尔有限公司版权所有。
The loss function is an important component in deep learning-based segmentation methods. Over the past five years, many loss functions have been proposed for various segmentation tasks. However, a systematic study of the utility of these loss functions is missing. In this paper, we present a comprehensive review of segmentation loss functions in an organized manner. We also conduct the first large-scale analysis of 20 general loss functions on four typical 3D segmentation tasks involving six public datasets from 10+ medical centers. The results show that none of the losses can consistently achieve the best performance on the four segmentation tasks, but compound loss functions (e.g. Dice with TopK loss, focal loss, Hausdorff distance loss, and boundary loss) are the most robust losses. Our code and segmentation results are publicly available and can serve as a loss function benchmark. We hope this work will also provide insights on new loss function development for the community. (c) 2021 Elsevier B.V. All rights reserved.