期刊
IEEE TRANSACTIONS ON MEDICAL IMAGING
卷 41, 期 4, 页码 965-976出版社
IEEE-INST ELECTRICAL ELECTRONICS ENGINEERS INC
DOI: 10.1109/TMI.2021.3130469
关键词
Image segmentation; Three-dimensional displays; Feature extraction; Computer architecture; Tumors; Convolution; Medical diagnostic imaging; Medical image segmentation; deep learning; overcomplete representations
类别
资金
- NSF [1910141]
- Direct For Computer & Info Scie & Enginr
- Div Of Information & Intelligent Systems [1910141] Funding Source: National Science Foundation
Most medical image segmentation methods use U-Net or its variants, but they often struggle with detecting smaller structures and accurately segmenting boundary regions. To address this issue, we propose KiU-Net, a convolutional architecture that combines overcomplete projection and U-Net to capture both high-level features and fine details. We demonstrate through experiments on multiple datasets that KiU-Net achieves good performance with fewer parameters and faster convergence.
Most methods for medical image segmentation use U-Net or its variants as they have been successful in most of the applications. After a detailed analysis of these traditional encoder-decoder based approaches, we observed that they perform poorly in detecting smaller structures and are unable to segment boundary regions precisely. This issue can be attributed to the increase in receptive field size as we go deeper into the encoder. The extra focus on learning high level features causes U-Net based approaches to learn less information about low-level features which are crucial for detecting small structures. To overcome this issue, we propose using an overcomplete convolutional architecture where we project the input image into a higher dimension such that we constrain the receptive field from increasing in the deep layers of the network. We design a new architecture for im- age segmentation- KiU-Net which has two branches: (1) an overcomplete convolutional network Kite-Net which learns to capture fine details and accurate edges of the input, and (2) U-Net which learns high level features. Furthermore, we also propose KiU-Net 3D which is a 3D convolutional architecture for volumetric segmentation. We perform a detailed study of KiU-Net by performing experiments on five different datasets covering various image modalities. We achieve a good performance with an additional benefit of fewer parameters and faster convergence. We also demonstrate that the extensions of KiU-Net based on residual blocks and dense blocks result in further performance improvements. Code: https://github.com/jeya-maria-jose/KiU-Net-pytorch
作者
我是这篇论文的作者
点击您的名字以认领此论文并将其添加到您的个人资料中。
推荐
暂无数据