Semi-Supervised Brain Lesion Segmentation with an Adapted Mean Teacher Model (2019-03-04T00:00:00.000000Z)

TL;DR

This work proposes a semi-supervised learning (SSL) approach to brain lesion segmentation, where unannotated data is incorporated into the training of CNNs and outperforms competing SSL-based methods on ischemic stroke lesion segmentsation.

Abstract

Automated brain lesion segmentation provides valuable information for the analysis and intervention of patients. In particular, methods that are based on convolutional neural networks (CNNs) have achieved state-of-the-art segmentation performance. However, CNNs usually require a decent amount of annotated data, which may be costly and time-consuming to obtain. Since unannotated data is generally abundant, it is desirable to use unannotated data to improve the segmentation performance for CNNs when limited annotated data is available. In this work, we propose a semi-supervised learning (SSL) approach to brain lesion segmentation, where unannotated data is incorporated into the training of CNNs. We adapt the mean teacher model, which is originally developed for SSL-based image classification, for brain lesion segmentation. Assuming that the network should produce consistent outputs for similar inputs, a loss of segmentation consistency is designed and integrated into a self-ensembling framework. Self-ensembling exploits the information in the intermediate training steps, and the ensemble prediction based on the information can be closer to the correct result than the single latest model. To exploit such information, we build a student model and a teacher model, which share the same CNN architecture for segmentation. The student and teacher models are updated alternately. At each step, the student model learns from the teacher model by minimizing the weighted sum of the segmentation loss computed from annotated data and the segmentation consistency loss between the teacher and student models computed from unannotated data. Then, the teacher model is updated by combining the updated student model with the historical information of teacher models using an exponential moving average strategy. For demonstration, the proposed approach was evaluated on ischemic stroke lesion segmentation. Results indicate that the proposed method improves stroke lesion segmentation with the incorporation of unannotated data and outperforms competing SSL-based methods.

Semi-Supervised Brain Lesion Segmentation with an Adapted Mean Teacher Model

TL;DR

Abstract

Authors

References17 items

Joint Segmentation of Intracerebral Hemorrhage and Infarct from Non-Contrast CT Images of Post-treatment Acute Ischemic Stroke Patients

Deep Adversarial Networks for Biomedical Image Segmentation Utilizing Unannotated Images

Semi-supervised Deep Learning for Fully Convolutional Networks

Weight-averaged consistency targets improve semi-supervised deep learning results

A deep learning model integrating FCNNs and CRFs for brain tumor segmentation

Unsupervised domain adaptation in brain lesion segmentation with adversarial networks

Temporal Ensembling for Semi-Supervised Learning

3D U-Net: Learning Dense Volumetric Segmentation from Sparse Annotation

Efficient multi‐scale 3D CNN with fully connected CRF for accurate brain lesion segmentation

Brain Tumor Segmentation Using Convolutional Neural Networks in MRI Images

Brain tumor segmentation with Deep Neural Networks

Dipy, a library for the analysis of diffusion MRI data

Local active information storage as a tool to understand distributed neural information processing

Dropout Training as Adaptive Regularization

A brief introduction to weakly supervised learning

ISLES 2015 ‐ A public evaluation benchmark for ischemic stroke lesion segmentation from multispectral MRI

Lecture 6.5-RMSProp, COURSERA: Neural networks for machine learning