Revisiting Computer-Aided Tuberculosis Diagnosis

TL;DR

This paper introduces a large-scale TB X-ray dataset, proposes a novel SymFormer model with SymAttention and SPE for improved TB detection, and establishes a benchmark for future research in computer-aided TB diagnosis.

Contribution

The paper presents the TBX11K dataset, a new baseline model SymFormer with innovative attention mechanisms, and a comprehensive benchmark for TB detection in chest X-rays.

Findings

SymFormer achieves state-of-the-art results on TBX11K.

The dataset enables training of high-quality TB detectors.

Benchmarking promotes future research in CTD.

Abstract

Tuberculosis (TB) is a major global health threat, causing millions of deaths annually. Although early diagnosis and treatment can greatly improve the chances of survival, it remains a major challenge, especially in developing countries. Recently, computer-aided tuberculosis diagnosis (CTD) using deep learning has shown promise, but progress is hindered by limited training data. To address this, we establish a large-scale dataset, namely the Tuberculosis X-ray (TBX11K) dataset, which contains 11,200 chest X-ray (CXR) images with corresponding bounding box annotations for TB areas. This dataset enables the training of sophisticated detectors for high-quality CTD. Furthermore, we propose a strong baseline, SymFormer, for simultaneous CXR image classification and TB infection area detection. SymFormer incorporates Symmetric Search Attention (SymAttention) to tackle the bilateral symmetry property of CXR images for learning discriminative features. Since CXR images may not strictly adhere to the bilateral symmetry property, we also propose Symmetric Positional Encoding (SPE) to facilitate SymAttention through feature recalibration. To promote future research on CTD, we build a benchmark by introducing evaluation metrics, evaluating baseline models reformed from existing detectors, and running an online challenge. Experiments show that SymFormer achieves state-of-the-art performance on the TBX11K dataset. The data, code, and models will be released at https://github.com/yun-liu/Tuberculosis.