Tuberculosis (TB) Chest X-ray Dataset for Automated Classification (4,200 Images)

This dataset contains a curated collection of 4,200 frontal chest X-ray (CXR) images designed for the automated detection and binary classification of pulmonary Tuberculosis (TB). The data is categorized into two classes: TB-positive (700 samples) and Normal (healthy) cases (3500 samples). This specific 4,200-image collection was utilized for the training, validation, and testing of computationally efficient hybrid deep learning architectures (specifically Vision Transformers) aimed at resource-constrained clinical environments. Origin & Attribution: This data is a specific subset/split derived from the larger public Tuberculosis (TB) Chest X-ray Database (initially curated by researchers from Qatar University and the University of Dhaka on Kaggle). The underlying images aggregate radiological scans from well-known public health sources, including: The National Library of Medicine (NLM) Montgomery and Shenzhen datasets The National Institute of Allergy and Infectious Diseases (NIAID) TB Portal The RSNA Pneumonia Detection Challenge dataset Structure & Format: To bypass file limits and preserve the internal folder structure, the 4,200 images have been compressed into a single .zip archive. The images are suitable for immediate preprocessing and integration into standard computer vision and deep learning pipelines.