Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

DiCoM -- Diverse Concept Modeling towards Enhancing Generalizability in Chest X-Ray Studies

Abhijeet Parida, Daniel Capellan-Martin, Sara Atito, Muhammad Awais, Maria J. Ledesma-Carbayo, Marius G. Linguraru, Syed Muhammad Anwar

TL;DR

DiCoM addresses the challenge of generalizable chest X-ray analysis under limited labeled data by a self-supervised, vision-transformer based framework that learns diverse concepts through a student–teacher setup and group masked modeling. It jointly optimizes reconstruction, token-level, and image-level pseudo-label losses to produce robust representations transferable to seen, unseen, and distribution-shifted tasks, including classification and segmentation. Across multiple public and private datasets, DiCoM consistently outperforms fully supervised baselines and other SSL methods, while demonstrating faster convergence and improved performance on pediatric and other out-of-distribution data. These properties position DiCoM as a strong candidate for a chest radiography foundation model with practical clinical impact and scalability.

Abstract

Chest X-Ray (CXR) is a widely used clinical imaging modality and has a pivotal role in the diagnosis and prognosis of various lung and heart related conditions. Conventional automated clinical diagnostic tool design strategies relying on radiology reads and supervised learning, entail the cumbersome requirement of high quality annotated training data. To address this challenge, self-supervised pre-training has proven to outperform supervised pre-training in numerous downstream vision tasks, representing a significant breakthrough in the field. However, medical imaging pre-training significantly differs from pre-training with natural images (e.g., ImageNet) due to unique attributes of clinical images. In this context, we introduce Diverse Concept Modeling (DiCoM), a novel self-supervised training paradigm that leverages a student teacher framework for learning diverse concepts and hence effective representation of the CXR data. Hence, expanding beyond merely modeling a single primary label within an image, instead, effectively harnessing the information from all the concepts inherent in the CXR. The pre-trained model is subsequently fine-tuned to address diverse domain-specific tasks. Our proposed paradigm consistently demonstrates robust performance across multiple downstream tasks on multiple datasets, highlighting the success and generalizability of the pre-training strategy. To establish the efficacy of our methods we analyze both the power of learned representations and the speed of convergence (SoC) of our models. For diverse data and tasks, DiCoM is able to achieve in most cases better results compared to other state-of-the-art pre-training strategies. This when combined with the higher SoC and generalization capabilities positions DiCoM to be established as a foundation model for CXRs, a widely used imaging modality.

DiCoM -- Diverse Concept Modeling towards Enhancing Generalizability in Chest X-Ray Studies

TL;DR

DiCoM addresses the challenge of generalizable chest X-ray analysis under limited labeled data by a self-supervised, vision-transformer based framework that learns diverse concepts through a student–teacher setup and group masked modeling. It jointly optimizes reconstruction, token-level, and image-level pseudo-label losses to produce robust representations transferable to seen, unseen, and distribution-shifted tasks, including classification and segmentation. Across multiple public and private datasets, DiCoM consistently outperforms fully supervised baselines and other SSL methods, while demonstrating faster convergence and improved performance on pediatric and other out-of-distribution data. These properties position DiCoM as a strong candidate for a chest radiography foundation model with practical clinical impact and scalability.

Abstract

Chest X-Ray (CXR) is a widely used clinical imaging modality and has a pivotal role in the diagnosis and prognosis of various lung and heart related conditions. Conventional automated clinical diagnostic tool design strategies relying on radiology reads and supervised learning, entail the cumbersome requirement of high quality annotated training data. To address this challenge, self-supervised pre-training has proven to outperform supervised pre-training in numerous downstream vision tasks, representing a significant breakthrough in the field. However, medical imaging pre-training significantly differs from pre-training with natural images (e.g., ImageNet) due to unique attributes of clinical images. In this context, we introduce Diverse Concept Modeling (DiCoM), a novel self-supervised training paradigm that leverages a student teacher framework for learning diverse concepts and hence effective representation of the CXR data. Hence, expanding beyond merely modeling a single primary label within an image, instead, effectively harnessing the information from all the concepts inherent in the CXR. The pre-trained model is subsequently fine-tuned to address diverse domain-specific tasks. Our proposed paradigm consistently demonstrates robust performance across multiple downstream tasks on multiple datasets, highlighting the success and generalizability of the pre-training strategy. To establish the efficacy of our methods we analyze both the power of learned representations and the speed of convergence (SoC) of our models. For diverse data and tasks, DiCoM is able to achieve in most cases better results compared to other state-of-the-art pre-training strategies. This when combined with the higher SoC and generalization capabilities positions DiCoM to be established as a foundation model for CXRs, a widely used imaging modality.
Paper Structure (19 sections, 6 equations, 3 figures, 6 tables)

This paper contains 19 sections, 6 equations, 3 figures, 6 tables.

Table of Contents

  1. Introduction
  2. Related Works
  3. Self-Supervised Pre-training for Medical Imaging
  4. Preliminary
  5. Local and Global Pseudo-Label Learning
  6. Putting Together the DiCoM Framework
  7. Experimental Results
  8. Datasets
  9. Evaluation Strategies
  10. Binary Classification
  11. Seen Distribution during Downstream Tuning
  12. Unseen Distribution during Downstream Tuning
  13. Shifted Distribution during Downstream Tuning
  14. Multi-class Classification
  15. Segmentation: Architecture Shift
  16. ...and 4 more sections

Figures (3)

  • Figure 1: Diverse Concept Modeling (DiCoM) framework for self-supervised pre-training of vision transformers for chest X-ray image analysis.
  • Figure 2: Speed of Convergence (SoC) for various methods. Higher SoC indicates better performance in low resource settings.
  • Figure 3: Performance summary of the various pretraining strategies on different tasks. Higher area is an indicator of better overall performance. Abbreviations: clf, classification.