Validation of an AI-Driven Sputum Analysis Platform for Accurate and Reliable Tuberculosis Detection in South-Western Uganda.
- 1Department of Pathology, Mbarara University of Science and Technology (MUST), Mbarara, Uganda.
- 2SRF Research and Training Centers, Mbarara, Uganda.
- 3Department of Laboratory Services, Mbarara Regional Referral Hospital (MRRH), Mbarara, Uganda.
Abstract
Background
Ziehl-Neelsen (ZN) sputum smear microscopy remains an accessible method for detecting acid-fast bacilli (AFB), but its interpretation is dependent on specimen quality and trained microscopy personnel. Artificial intelligence (AI) may provide a microscopy reading aid that supports consistent identification of AFB-positive and AFB-negative smear images.Objective
To evaluate transfer-learning deep learning models for classification of AFB-positive and AFB-negative digitized ZN-stained sputum smear images generated in a Ugandan regional referral laboratory.Methods
A retrospective laboratory-based cross-sectional study used 512 archived digitized ZN-stained sputum smear images from Mbarara Regional Referral Hospital. The dataset contained 256 AFB-positive and 256 AFB-negative images and was partitioned by stratified sampling into training (n=358), validation (n=77), and an internal hold-out test set (n=77). ResNet-50 V2, DenseNet-121, and Swin Transformer-Tiny models were trained using ImageNet-pretrained weights in PyTorch. Performance was assessed using sensitivity, specificity, accuracy, positive predictive value (PPV), negative predictive value (NPV), F1-score, and ROC-AUC.Results
On the internal test set, Swin Transformer-Tiny achieved 97.40% accuracy, 97.37% sensitivity, 97.44% specificity, 97.37% PPV, 97.44% NPV, 97.37% F1-score, and ROC-AUC 0.9939. DenseNet-121 achieved 88.31% accuracy, 84.21% sensitivity, 92.31% specificity, and ROC-AUC 0.9730. ResNet-50 V2 achieved 85.71% accuracy, 81.58% sensitivity, 89.74% specificity, and ROC-AUC 0.9555. Wilson 95% confidence intervals were calculated for the threshold-based metrics.Conclusion
The models demonstrated promising internal performance for classification of microscopy-defined AFB status. Swin Transformer-Tiny had the highest observed performance, but the small, balanced, single-centre internal test set and use of routine microscopy labels rather than an independent microbiological reference standard limit clinical interpretation. The platform should therefore be considered an AI-assisted microscopy reading aid rather than an autonomous tuberculosis diagnostic system. Prospective, multicentre evaluation using patient/specimen-level separation, independent microbiological reference standards, and multiple blinded readers is required before clinical deployment.
Related
- Development and Validation of a Point-of-Care Triage Scorecard to Enhance Tuberculosis Case Detection During Active Community Screening in Yogyakarta, Indonesia
- Cross-Cultural Adaptation and Psychometric Validation of the Arabic Duruöz Hand Index: A Reliable Patient-Reported Measure for Hand Disability in Arabic-Speaking Populations
- Reliable Detection, Unreliable Clinical Decision-making: Artificial Intelligence Models in Burn Assessment
- Caputo fractional-order differential equations for tuberculosis transmission and relapse dynamics with empirical validation
- An Artificial Intelligence Supervision Program to Improve Video-Observed Therapy for Managing Tuberculosis Treatment: Development and Validation Study
BibTeX
@article{tumwesigye2026validation,
title = {{Validation of an AI-Driven Sputum Analysis Platform for Accurate and Reliable Tuberculosis Detection in South-Western Uganda.}},
author = {Habert Tumwesigye and Daniel Atwine and Raymond Atwine and Abraham Birungi and Caroline Birungi and Esther Munguciada and Phionah Nabimanya and Joseph Akandwanaho and Benard Ochama and Yekosani Mitala},
journal = {Medical devices (Auckland, N.Z.)},
year = {2026},
month = oct,
doi = {10.2147/mder.s639007},
url = {https://europepmc.org/article/MED/42899212},
note = {CC BY-NC},
}
RIS
TY - JOUR
TI - Validation of an AI-Driven Sputum Analysis Platform for Accurate and Reliable Tuberculosis Detection in South-Western Uganda.
AU - Habert Tumwesigye
AU - Daniel Atwine
AU - Raymond Atwine
AU - Abraham Birungi
AU - Caroline Birungi
AU - Esther Munguciada
AU - Phionah Nabimanya
AU - Joseph Akandwanaho
AU - Benard Ochama
AU - Yekosani Mitala
PY - 2026
DA - 2026/10/15
JO - Medical devices (Auckland, N.Z.)
DO - 10.2147/mder.s639007
UR - https://europepmc.org/article/MED/42899212
AB - Background Ziehl-Neelsen (ZN) sputum smear microscopy remains an accessible method for detecting acid-fast bacilli (AFB), but its interpretation is dependent on specimen quality and trained microscopy personnel. Artificial intelligence (AI) may provide a microscopy reading aid that supports consistent identification of AFB-positive and AFB-negative smear images.Objective To evaluate transfer-learning deep learning models for classification of AFB-positive and AFB-negative digitized ZN-stained sputum smear images generated in a Ugandan regional referral laboratory.Methods A retrospective laboratory-based cross-sectional study used 512 archived digitized ZN-stained sputum smear images from Mbarara Regional Referral Hospital. The dataset contained 256 AFB-positive and 256 AFB-negative images and was partitioned by stratified sampling into training (n=358), validation (n=77), and an internal hold-out test set (n=77). ResNet-50 V2, DenseNet-121, and Swin Transformer-Tiny models were trained using ImageNet-pretrained weights in PyTorch. Performance was assessed using sensitivity, specificity, accuracy, positive predictive value (PPV), negative predictive value (NPV), F1-score, and ROC-AUC.Results On the internal test set, Swin Transformer-Tiny achieved 97.40% accuracy, 97.37% sensitivity, 97.44% specificity, 97.37% PPV, 97.44% NPV, 97.37% F1-score, and ROC-AUC 0.9939. DenseNet-121 achieved 88.31% accuracy, 84.21% sensitivity, 92.31% specificity, and ROC-AUC 0.9730. ResNet-50 V2 achieved 85.71% accuracy, 81.58% sensitivity, 89.74% specificity, and ROC-AUC 0.9555. Wilson 95% confidence intervals were calculated for the threshold-based metrics.Conclusion The models demonstrated promising internal performance for classification of microscopy-defined AFB status. Swin Transformer-Tiny had the highest observed performance, but the small, balanced, single-centre internal test set and use of routine microscopy labels rather than an independent microbiological reference standard limit clinical interpretation. The platform should therefore be considered an AI-assisted microscopy reading aid rather than an autonomous tuberculosis diagnostic system. Prospective, multicentre evaluation using patient/specimen-level separation, independent microbiological reference standards, and multiple blinded readers is required before clinical deployment.
SN - 1179-1470
ER -