nnU-Net: a self-configuring method for deep learning-based biomedical image segmentation
nnU-Net: a self-configuring method for deep learning-based biomedical image segmentation
复制标题
nnU-Net：一种基于深度学习的生物医学图像分割自配置方法
DOI：
10.1038/s41592-020-01008-z
复制
发表时间：
2020-12-07
期刊：
影响因子：
48
通讯作者：
Maier-Hein, Klaus H.
中科院分区：
文献类型：
--
作者：
Isensee, Fabian;Jaeger, Paul F.;Maier-Hein, Klaus H.
NNU-Net是一种基于深度学习的图像分割方法，它能够自动配置自身来完成各种生物和医学图像分割任务。NNU-Net作为一种开箱即用的工具提供了最先进的性能。生物医学成像是科学发现的驱动力和医疗保健的核心组成部分，并受到深度学习领域的刺激。虽然语义分割算法能够在许多应用中实现图像分析和量化，但各个专门解决方案的设计不是平凡的，并且高度依赖于数据集的属性和硬件条件。我们开发了NNU-Net，这是一种基于深度学习的分割方法，可以自动配置自身，包括任何新任务的预处理、网络结构、训练和后处理。这一过程中的关键设计选择被建模为一组固定的参数、相互依赖的规则和经验决策。在没有人工干预的情况下，NNU-Net超过了大多数现有的方法，包括在国际生物医学分割比赛中使用的23个公共数据集上的高度专业化的解决方案。我们将NNU-Net作为一种开箱即用的工具公开提供，通过除了标准网络培训之外既不需要专业知识也不需要计算资源，使广大受众可以访问最先进的细分。
nnU-Net is a deep learning-based image segmentation method that automatically configures itself for diverse biological and medical image segmentation tasks. nnU-Net offers state-of-the-art performance as an out-of-the-box tool.Biomedical imaging is a driver of scientific discovery and a core component of medical care and is being stimulated by the field of deep learning. While semantic segmentation algorithms enable image analysis and quantification in many applications, the design of respective specialized solutions is non-trivial and highly dependent on dataset properties and hardware conditions. We developed nnU-Net, a deep learning-based segmentation method that automatically configures itself, including preprocessing, network architecture, training and post-processing for any new task. The key design choices in this process are modeled as a set of fixed parameters, interdependent rules and empirical decisions. Without manual intervention, nnU-Net surpasses most existing approaches, including highly specialized solutions on 23 public datasets used in international biomedical segmentation competitions. We make nnU-Net publicly available as an out-of-the-box tool, rendering state-of-the-art segmentation accessible to a broad audience by requiring neither expert knowledge nor computing resources beyond standard network training.