ACTION++: Improving Semi-supervised Medical Image Segmentation with Adaptive Anatomical Contrast (2023-04-05T00:00:00.000000Z)

TL;DR

This work proposes an adaptive supervised contrastive loss, where it is argued that blindly adopting a constant temperature τ in the contrastive loss on long-tailed medical data is not optimal, and proposes to use a dynamic τ via a simple cosine schedule to yield better separation between majority and minority classes.

Abstract

Medical data often exhibits long-tail distributions with heavy class imbalance, which naturally leads to difficulty in classifying the minority classes (i.e., boundary regions or rare objects). Recent work has significantly improved semi-supervised medical image segmentation in long-tailed scenarios by equipping them with unsupervised contrastive criteria. However, it remains unclear how well they will perform in the labeled portion of data where class distribution is also highly imbalanced. In this work, we present ACTION++, an improved contrastive learning framework with adaptive anatomical contrast for semi-supervised medical segmentation. Specifically, we propose an adaptive supervised contrastive loss, where we first compute the optimal locations of class centers uniformly distributed on the embedding space (i.e., off-line), and then perform online contrastive matching training by encouraging different class features to adaptively match these distinct and uniformly distributed class centers. Moreover, we argue that blindly adopting a constant temperature τ in the contrastive loss on long-tailed medical data is not optimal, and propose to use a dynamic τ via a simple cosine schedule to yield better separation between majority and minority classes. Empirically, we evaluate ACTION++ on ACDC and LA benchmarks and show that it achieves state-of-the-art across two semi-supervised settings. Theoretically, we analyze the performance of adaptive anatomical contrast and confirm its superiority in label efficiency.

Authors

References44 items

Implicit Anatomical Rendering for Medical Image Segmentation with Stochastic Experts

Temperature Schedules for Self-Supervised Contrastive Methods on Long-Tail Data

Rethinking Semi-Supervised Medical Image Segmentation: A Variance-Reduction Perspective

Information-guided pixel augmentation for pixel-wise contrastive learning

Mine yOur owN Anatomy: Revisiting Medical Image Segmentation With Extremely Limited Labels

Bootstrapping Semi-supervised Medical Image Segmentation with Anatomical-aware Contrastive Distillation

Incremental Learning Meets Transfer Learning: Application to Multi-site Prostate MRI Segmentation

SaR: Self-adaptive Refinement on Pseudo Labels for Multiclass-Imbalanced Semi-supervised Learning

Exploring Smoothness and Class-Separation for Semi-supervised Medical Image Segmentation

Class-Aware Adversarial Transformers for Medical Image Segmentation

Targeted Supervised Contrastive Learning for Long-Tailed Recognition

Towards the Generalization of Contrastive Self-Supervised Learning

Joint Semi-supervised and Active Learning for Segmentation of Gigapixel Pathology Images with Cost-Effective Labeling

SimCVD: Simple Contrastive Voxel-Wise Representation Distillation for Semi-Supervised Medical Image Segmentation

Interpretable Minority Synthesis for Imbalanced Classification

Can contrastive learning avoid shortcut solutions?

Momentum Contrastive Voxel-wise Representation Learning for Semi-supervised Volumetric Medical Image Segmentation

Semi-supervised Left Atrium Segmentation with Mutual Consistency Training

Dissecting Supervised Contrastive Learning

Analyzing Overfitting Under Class Imbalance in Neural Networks for Image Segmentation

Understanding the Behaviour of Contrastive Loss

Efficient Semi-supervised Gross Target Volume of Nasopharyngeal Carcinoma Segmentation via Uncertainty Rectified Pyramid Consistency

A global benchmark of algorithms for segmenting the left atrium from late gadolinium-enhanced cardiac magnetic resonance imaging

Semi-supervised Medical Image Segmentation through Dual-task Consistency

Unsupervised Wasserstein Distance Guided Domain Adaptation for 3D Multi-Domain Liver Segmentation

Shape-aware Semi-supervised 3D Semantic Segmentation for Medical Images

Contrastive learning of global and local features for medical image segmentation with limited annotations

Improved Baselines with Momentum Contrastive Learning

A Simple Framework for Contrastive Learning of Visual Representations

Decoupling Representation and Classifier for Long-Tailed Recognition

Uncertainty-aware Self-ensembling Model for Semi-supervised 3D Left Atrium Segmentation

Curriculum semi-supervised segmentation

Deep Learning Techniques for Automatic MRI Cardiac Multi-Structures Segmentation and Diagnosis: Is the Problem Solved?

Semi-Supervised 3D Abdominal Multi-Organ Segmentation Via Deep Multi-Planar Co-Training

Deep Co-Training for Semi-Supervised Image Recognition

SegAN: Adversarial Network with Multi-scale L1 Loss for Medical Image Segmentation

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

V-Net: Fully Convolutional Neural Networks for Volumetric Medical Image Segmentation

U-Net: Convolutional Networks for Biomedical Image Segmentation

The Psycho-Biology Of Language: AN INTRODUCTION TO DYNAMIC PHILOLOGY

Smoothed Adaptive Weighting for Imbalanced Semi-Supervised Learning: Improve Reliability Against Unknown Distribution Data

Generalizable semi-supervised learning strategies for multiple learning tasks using 1-d biomedical signals

Exploring Balanced Feature Spaces for Representation Learning

Semi-Supervised Learning (Chapelle, O. et al., Eds.; 2006) [Book reviews]