Published on Mon May 17 2021

Cross-Modality Brain Tumor Segmentation via Bidirectional Global-to-Local Unsupervised Domain Adaptation

Kelei He, Wen Ji, Tao Zhou, Zhuoyuan Li, Jing Huo, Xin Zhang, Yang Gao, Dinggang Shen, Bing Zhang, Junfeng Zhang

Accurate segmentation of brain tumors from multi-modal Magnetic Resonance images is essential in brain tumor diagnosis and treatment. Due to the existence of domain shifts among different modalities, the performance of networks decreases dramatically when training on one modality and performing on another. In this paper, we propose a novel Bidirectional Global-to-

0
0
0
Abstract

Accurate segmentation of brain tumors from multi-modal Magnetic Resonance (MR) images is essential in brain tumor diagnosis and treatment. However, due to the existence of domain shifts among different modalities, the performance of networks decreases dramatically when training on one modality and performing on another, e.g., train on T1 image while performing on T2 image, which is often required in clinical applications. This also prohibits a network from being trained on labeled data and then transferred to unlabeled data from a different domain. To overcome this, unsupervised domain adaptation (UDA) methods provide effective solutions to alleviate the domain shift between labeled source data and unlabeled target data. In this paper, we propose a novel Bidirectional Global-to-Local (BiGL) adaptation framework under a UDA scheme. Specifically, a bidirectional image synthesis and segmentation module is proposed to segment the brain tumor using the intermediate data distributions generated for the two domains, which includes an image-to-image translator and a shared-weighted segmentation network. Further, a global-to-local consistency learning module is proposed to build robust representation alignments in an integrated way. Extensive experiments on a multi-modal brain MR benchmark dataset demonstrate that the proposed method outperforms several state-of-the-art unsupervised domain adaptation methods by a large margin, while a comprehensive ablation study validates the effectiveness of each key component. The implementation code of our method will be released at \url{https://github.com/KeleiHe/BiGL}.

Mon May 11 2020
Computer Vision
An Inductive Transfer Learning Approach using Cycle-consistent Adversarial Domain Adaptation with Application to Brain Tumor Segmentation
The multi-modal nature of each annotated image renders it difficult to share the annotation label among diverse tasks. The proposed ITL approach can make significant contribution to the field of medical image analysis, as we develop a fundamental tool.
0
0
0
Thu Feb 06 2020
Computer Vision
Unsupervised Bidirectional Cross-Modality Adaptation via Deeply Synergistic Image and Feature Alignment for Medical Image Segmentation
Unsupervised domain adaptation has increasingly gained interest in medical image computing. Synergistic Image and Feature Alignment (SIFA) is a framework to effectively adapt a segmentation network to an unlabeled target domain.
0
0
0
Sun Sep 06 2020
Computer Vision
Unsupervised Wasserstein Distance Guided Domain Adaptation for 3D Multi-Domain Liver Segmentation
Deep neural networks have shown exceptional learning capability and generalizability in the source domain when massive labeled data is provided. However, the well-trained models often fail in the target domain due to the domain shift. To mitigate these issues, we utilize Wasserstein distance to learn more complete representation.
0
0
0
Mon Jun 08 2020
Computer Vision
Cross-Domain Segmentation with Adversarial Loss and Covariate Shift for Biomedical Imaging
The capacity of current segmentation techniques is only allow to work for an individual domain. The models that are capable of working on all modalities are essentially needed for a complete solution. The tests on CT and MRI liver data acquired inoutine clinical workflows show that the proposed model outperforms all other baseline with a large margin.
0
0
0
Thu Jun 17 2021
Computer Vision
Knowledge distillation from multi-modal to mono-modal segmentation networks
The joint use of multiple imaging modalities for medical image segmentation has been widely studied in recent years. Most of the time, only one modality is acquired. The proposed method is an adaptation of the generalized distillation framework.
0
0
0
Wed Jul 31 2019
Computer Vision
Unsupervised Domain Adaptation via Disentangled Representations: Application to Cross-Modality Liver Segmentation
A deep learning model trained on some labeled data from a certain source generally performs poorly on data from different target domains due to domain shifts. Unsupervised domain adaptation methods address this problem byleviating the domain shift between the labeled source data and the unlabeled target data.
0
0
0
Wed Sep 25 2019
Computer Vision
A Closer Look at Domain Shift for Deep Learning in Histopathology
Domain shift is a significant problem in histopathology. There can be large disparities in data characteristics of whole-slide images. This makes generalization of deep learning to unseen data difficult.
1
1
1
Sun Aug 21 2016
Computer Vision
VoxResNet: Deep Voxelwise Residual Networks for Volumetric Brain Segmentation
Deep residual learning with residual units for training very deep neural networks advanced the state-of-the-art performance on 2D image recognition tasks. How to leverage contextual representations for recognition tasks from volumetric data has not been well studied, especially in the field of medical image computing.
0
0
0
Wed May 10 2017
Computer Vision
Automatic Brain Tumor Detection and Segmentation Using U-Net Based Fully Convolutional Networks
A major challenge in brain tumor treatment planning and quantitative evaluating is determination of the tumor extent. Manual segmentation from 3D MRI volumes is a very time-consuming task and the performance is highly relied on operator's experience. We propose a fully autoimmune method for brain tumor segmentation.
0
0
0
Sat Nov 09 2019
Machine Learning
Deep learning for cardiac image segmentation: A review
Deep learning has become the most widely used approach for cardiac image segmentation in recent years. This paper provides a review of over 100 papers using deep learning. We discuss the challenges and limitations with current deep learning-based approaches and suggest potential directions for future research.
0
0
0
Tue Feb 11 2020
Computer Vision
Hi-Net: Hybrid-fusion Network for Multi-modal MR Image Synthesis
Magnetic resonance imaging (MRI) is a widely used neuroimaging technique that provides images of different contrasts (i.e., modalities) Fusing this multi-modal data has proven particularly effective for boosting model performance in many tasks. Medical image synthesis has been proposed as an effective solution to this.
0
0
0
Thu Jan 24 2019
Computer Vision
Synergistic Image and Feature Adaptation: Towards Cross-Modality Domain Adaptation for Medical Image Segmentation
This paper presents a novel unsupervised domain adaptation framework, called Synergistic Image and Feature Adaptation (SIFA) It aims to recover performance degradation when applying the neural networks to new testing domains.
0
0
0