AI-TB-ZN: An Annotated Ziehl–Neelsen-Stained Sputum Smear Image Dataset for Artificial Intelligence-Based Tuberculosis Detection in Southwestern Uganda

AI-TB-ZN: An Annotated Ziehl–Neelsen-Stained Sputum Smear Image Dataset for Artificial Intelligence-Based Tuberculosis Detection in Southwestern Uganda The AI-TB-ZN dataset is a publicly available collection of 512 anonymized Ziehl–Neelsen (ZN)-stained sputum smear microscopy images developed to support research in artificial intelligence (AI), machine learning (ML), deep learning (DL), computer vision, and computer-aided tuberculosis (TB) diagnosis. The dataset contains: 512 JPEG images 256 Acid-Fast Bacilli (AFB)-positive images 256 AFB-negative images Image-level expert annotations labels.csv metadata.csv README Dataset paper Data Availability Statement LICENSE (CC BY 4.0) CITATION.cff Image Acquisition Images were acquired from archived Ziehl–Neelsen-stained sputum smear slides collected during routine tuberculosis diagnosis in Southwestern Uganda. Image acquisition was performed using: Olympus BX41 light microscope Infinity Lite digital microscope camera Samsung Galaxy A16 smartphone (used for a subset of images) ×100 magnification All images were anonymized before publication and contain no patient-identifying information. Dataset Applications The dataset is intended for: Artificial intelligence Machine learning Deep learning Medical image analysis Computer vision Explainable AI Transfer learning Tuberculosis diagnosis Educational purposes Benchmarking diagnostic algorithms Funding This work was supported by the Faculty of Health Sciences Seed Grant, Mbarara University of Science and Technology (MUST), Uganda. Ethical Approval Ethical approval was obtained from the Mbarara University of Science and Technology Research Ethics Committee (MUST-REC; Approval No. MUST-2025-618). Administrative clearance was granted by Mbarara Regional Referral Hospital (MRRH). The released dataset contains only anonymized images. License This dataset is distributed under the Creative Commons Attribution 4.0 International (CC BY 4.0) License.