Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Computer-Aided Diagnosis of Thoracic Diseases in Chest X-rays using hybrid CNN-Transformer Architecture

Sonit Singh

TL;DR

The paper tackles automated, multi-label thoracic disease detection in chest X-rays using a transformer-augmented CNN architecture. It introduces SA-DenseNet121, a DenseNet-121 backbone enhanced with multi-head self-attention to capture both local and global features. Evaluations across ChestX-ray14, CheXpert, MIMIC-CXR-JPG, and IU-CXR show improved AUC-ROC over strong baselines, particularly for smaller lesions, indicating better localization and discrimination. The approach has practical potential to augment radiologists by providing rapid second opinions and highlighted regions, though challenges such as label noise and dataset heterogeneity remain to be addressed.

Abstract

Medical imaging has been used for diagnosis of various conditions, making it one of the most powerful resources for effective patient care. Due to widespread availability, low cost, and low radiation, chest X-ray is one of the most sought after radiology examination for the diagnosis of various thoracic diseases. Due to advancements in medical imaging technologies and increasing patient load, current radiology workflow faces various challenges including increasing backlogs, working long hours, and increase in diagnostic errors. An automated computer-aided diagnosis system that can interpret chest X-rays to augment radiologists by providing actionable insights has potential to provide second opinion to radiologists, highlight relevant regions in the image, in turn expediting clinical workflow, reducing diagnostic errors, and improving patient care. In this study, we applied a novel architecture augmenting the DenseNet121 Convolutional Neural Network (CNN) with multi-head self-attention mechanism using transformer, namely SA-DenseNet121, that can identify multiple thoracic diseases in chest X-rays. We conducted experiments on four of the largest chest X-ray datasets, namely, ChestX-ray14, CheXpert, MIMIC-CXR-JPG, and IU-CXR. Experimental results in terms of area under the receiver operating characteristics (AUC-ROC) shows that augmenting CNN with self-attention has potential in diagnosing different thoracic diseases from chest X-rays. The proposed methodology has the potential to support the reading workflow, improve efficiency, and reduce diagnostic errors.

Computer-Aided Diagnosis of Thoracic Diseases in Chest X-rays using hybrid CNN-Transformer Architecture

TL;DR

The paper tackles automated, multi-label thoracic disease detection in chest X-rays using a transformer-augmented CNN architecture. It introduces SA-DenseNet121, a DenseNet-121 backbone enhanced with multi-head self-attention to capture both local and global features. Evaluations across ChestX-ray14, CheXpert, MIMIC-CXR-JPG, and IU-CXR show improved AUC-ROC over strong baselines, particularly for smaller lesions, indicating better localization and discrimination. The approach has practical potential to augment radiologists by providing rapid second opinions and highlighted regions, though challenges such as label noise and dataset heterogeneity remain to be addressed.

Abstract

Medical imaging has been used for diagnosis of various conditions, making it one of the most powerful resources for effective patient care. Due to widespread availability, low cost, and low radiation, chest X-ray is one of the most sought after radiology examination for the diagnosis of various thoracic diseases. Due to advancements in medical imaging technologies and increasing patient load, current radiology workflow faces various challenges including increasing backlogs, working long hours, and increase in diagnostic errors. An automated computer-aided diagnosis system that can interpret chest X-rays to augment radiologists by providing actionable insights has potential to provide second opinion to radiologists, highlight relevant regions in the image, in turn expediting clinical workflow, reducing diagnostic errors, and improving patient care. In this study, we applied a novel architecture augmenting the DenseNet121 Convolutional Neural Network (CNN) with multi-head self-attention mechanism using transformer, namely SA-DenseNet121, that can identify multiple thoracic diseases in chest X-rays. We conducted experiments on four of the largest chest X-ray datasets, namely, ChestX-ray14, CheXpert, MIMIC-CXR-JPG, and IU-CXR. Experimental results in terms of area under the receiver operating characteristics (AUC-ROC) shows that augmenting CNN with self-attention has potential in diagnosing different thoracic diseases from chest X-rays. The proposed methodology has the potential to support the reading workflow, improve efficiency, and reduce diagnostic errors.
Paper Structure (26 sections, 6 equations, 13 figures, 13 tables)

This paper contains 26 sections, 6 equations, 13 figures, 13 tables.

Table of Contents

  1. Introduction
  2. Related Work
  3. Chest X-ray datasets
  4. Thoracic diseases detection
  5. Attention models in thoracic disease detection
  6. Methodology
  7. Convolutional Neural Networks
  8. Multi-head self-attention
  9. Data augmentation
  10. Transfer learning
  11. CheXpert-An open-source radiology report labeler
  12. Handling uncertain labels by CheXpert labeler
  13. Experimental setup
  14. Problem definition
  15. Datasets
  16. ...and 11 more sections

Figures (13)

  • Figure 1: Block diagram of multi-head self-attention based convolutional network for thoracic diseases identification.
  • Figure 2: Randomly sampled chest X-ray having Cardiomegaly from each of the four datasets: ChestX-ray14, CheXpert, MIMIC-CXR-JPG, and IU-CXR.
  • Figure 3: DenseNet-121 model architecture Ho:gwak:2019:multi-feature_integration.
  • Figure 4: Transformer architecture, multi-head attention block, and scaled dot product attention vaswani:2017:attention_is_all_you_need.
  • Figure 5: CheXpert labeler output on various observations on a sample radiology report.
  • ...and 8 more figures