Reliable tuberculosis detection from chest X-ray images requires both discriminative representation learning and evaluation protocols that minimize data leakage and source-related bias. This study proposes TBFuseFormer, a dual-branch framework that integrates hierarchical local features from Swin Transformer V2 with global semantic representations from a pretrained DINOv2 encoder. The two feature streams are combined through feature concatenation and channel attention-guided fusion for binary classification of normal and TB-positive chest radiographs. Experiments were conducted on a curated multi-source dataset containing 7698 images, comprising 4220 normal and 3478 TB-positive cases. Duplicate and near-duplicate images were removed before partitioning, and the final dataset was divided into training, validation, and held-out test sets of 5389, 1155, and 1154 images, respectively. To ensure fair comparison, all nine evaluated architectures underwent independent model-specific hyperparameter optimization. TBFuseFormer achieved a five-fold cross-validation accuracy of 99.51 ± 0.12% after optimization and a held-out test accuracy of 99.39%, outperforming tuned Swin Transformer V2 Base, ConvNeXt V2 Base, and standalone DINOv2 ViT-B/14. Leave-one-source-out evaluation further demonstrated strong unseen-source generalization, with a mean accuracy of 95.23%, balanced accuracy of 94.36%, F1-score of 93.39%, Matthews correlation coefficient of 0.864, and AUROC of 0.9836. These findings indicate that TBFuseFormer provides strong leakage-aware classification performance, robust cross-source discrimination, quantitatively supported localization faithfulness, and practical deployment options, although independent multicenter and prospective validation remains necessary.
Rahim, Abdur, Shehan, Md. Sazzad Hossain, Islam, Mazharul et al. Leakage aware tuberculosis detection from chest X-ray images using TBFuseFormer with Swin Transformer V2 and pretrained DINOv2 feature fusion. // Discover Analytics. – 2026. – DOI: 10.1007/s44257-026-00092-3. – URL: http://link.springer.com/openurl/fulltext?id=doi:10.1007/s44257-026-00092-3
Rahim, Abdur, Shehan, Md. Sazzad Hossain, Islam, Mazharul, Rahman, Md. Mizanur, Rana, Md. Sumon, Islam, Md. Azharul, Alam, Md. Fakrul, Jahan, Sumayya, Raihan, & Dinar Ibn (2026). Leakage aware tuberculosis detection from chest X-ray images using TBFuseFormer with Swin Transformer V2 and pretrained DINOv2 feature fusion. Discover Analytics. https://doi.org/10.1007/s44257-026-00092-3
Rahim, et al. “Leakage aware tuberculosis detection from chest X-ray images using TBFuseFormer with Swin Transformer V2 and pretrained DINOv2 feature fusion.” Discover Analytics, 2026. doi:10.1007/s44257-026-00092-3.
@article{Rahim2026,
title = {Leakage aware tuberculosis detection from chest X-ray images using TBFuseFormer with Swin Transformer V2 and pretrained DINOv2 feature fusion},
author = {Rahim and Abdur and Shehan and Hossain, Md. Sazzad and Islam and Mazharul and Rahman and Mizanur, Md. and Rana and Sumon, Md. and Islam and Azharul, Md. and Alam and Fakrul, Md. and Jahan and Sumayya and Raihan and Ibn, Dinar},
journal = {Discover Analytics},
year = {2026},
doi = {10.1007/s44257-026-00092-3},
url = {http://link.springer.com/openurl/fulltext?id=doi:10.1007/s44257-026-00092-3}
}
The diffusion of Artificial Intelligence (AI) is reshaping the competitive architecture of small and medium enterprises (SMEs), yet the extent to which AI adoption is associated with measurable competitive advantage among...
This study draws on Uses and Gratifications Theory (UGT), the MAIN model, and prior OTT service research to examine key factors associated with consumer satisfaction with OTT SVOD platforms, focusing on entertainment, escape, modality,...
Research has shown that entrepreneurial ecosystems (EEs) shape universities’ entrepreneurial activities. However, limited attention has been paid to how university entrepreneurial ecosystems (UEEs) support graduate start-up creation...
This study investigates the safe-haven resilience of Gold, the US dollar (USD), and Bitcoin against four systemic tail risks—stock market volatility (VIX), oil market volatility (OVX), geopolitical risk (GPR), and climate risk—over the...
NA. Tanzania’s sisal exports are an important source of foreign exchange and rural income, but their value is volatile, reflecting global demand cycles, competition from synthetic fibres, climatic variability and domestic supply...