Automatic Segmentation of Organs-at-Risk from Head-and-Neck CT using Separable Convolutional Neural Network with Hard-Region-Weighted Loss
This work aims to improve the accuracy and provide uncertainty estimation for OAR segmentation in Head-and-Neck CT scans, which is critical for radiation therapy planning for NPC patients. This is an incremental improvement in a specific medical imaging domain.
This paper addresses the challenge of segmenting Organs-at-Risk (OARs) from Head-and-Neck CT scans for Nasopharyngeal Carcinoma (NPC) treatment planning. The authors propose a novel framework that includes a Segmental Linear Function for intensity transformation, a 2.5D network (3D-SepNet) to handle anisotropic slice spacing, and a voxel-level hard-region-weighted loss (ATH) for improved accuracy.
Nasopharyngeal Carcinoma (NPC) is a leading form of Head-and-Neck (HAN) cancer in the Arctic, China, Southeast Asia, and the Middle East/North Africa. Accurate segmentation of Organs-at-Risk (OAR) from Computed Tomography (CT) images with uncertainty information is critical for effective planning of radiation therapy for NPC treatment. Despite the stateof-the-art performance achieved by Convolutional Neural Networks (CNNs) for automatic segmentation of OARs, existing methods do not provide uncertainty estimation of the segmentation results for treatment planning, and their accuracy is still limited by several factors, including the low contrast of soft tissues in CT, highly imbalanced sizes of OARs and large inter-slice spacing. To address these problems, we propose a novel framework for accurate OAR segmentation with reliable uncertainty estimation. First, we propose a Segmental Linear Function (SLF) to transform the intensity of CT images to make multiple organs more distinguishable than existing methods based on a simple window width/level that often gives a better visibility of one organ while hiding the others. Second, to deal with the large inter-slice spacing, we introduce a novel 2.5D network (named as 3D-SepNet) specially designed for dealing with clinic HAN CT scans with anisotropic spacing. Thirdly, existing hardness-aware loss function often deal with class-level hardness, but our proposed attention to hard voxels (ATH) uses a voxel-level hardness strategy, which is more suitable to dealing with some hard regions despite that its corresponding class may be easy. Our code is now available at https://github.com/HiLab-git/SepNet.