EUVIMEDEUVIMED entdecken
Crossref

Tuberculosis classification from chest X-rays using hybrid CNN-transformer architecture leveraging LSTM feature selection

Saida Sarra Boudouh

IET Conference Proceedings · 01.08.2026 · 2025 (63) · 146-150

DOI: 10.1049/icp.2026.2352 ↗

Kurz gefasst

Fachartikel in der Zeitschrift IET Conference Proceedings, erschienen am 01.08.2026. Verfasst von Saida Sarra Boudouh.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Tuberculosis remains one of the world’s deadliest infectious diseases, posing a significant public health challenge, especially in low-resource settings. This paper presents a hybrid deep learning framework for automated Tuberculosis detection using chest X-ray images. The proposed architecture integrates a Vision Transformer encoder and a partially frozen Xception convolutional neural network to extract both global contextual and local discriminative features. We selectively activated deeper blocks from the middle and existing flow of the Xception model to focus on high-level representations while minimizing noise from earlier layers. The extracted features from both branches were concatenated and passed through a Long Short-Term Memory network for effective feature selection. This strategy allowed the model to retain only the most relevant information for classification. Evaluation on a balanced dataset of 7,000 images demonstrated the effectiveness of our approach, achieving an accuracy of 99.93% and a recall of 99.84%. These results highlight the potential of combining heterogeneous deep features and temporal feature selection for robust and precise Tuberculosis screening in medical imaging.

Zitieren

APA

Boudouh, S. S. (2026). Tuberculosis classification from chest X-rays using hybrid CNN-transformer architecture leveraging LSTM feature selection. IET Conference Proceedings, 2025(63), 146-150. https://doi.org/10.1049/icp.2026.2352

Vancouver

Boudouh SS. Tuberculosis classification from chest X-rays using hybrid CNN-transformer architecture leveraging LSTM feature selection. IET Conference Proceedings. 2026;2025(63):146-150. doi:10.1049/icp.2026.2352

BibTeX
@article{Boudouh2026,
  author = {Boudouh, Saida Sarra},
  title = {Tuberculosis classification from chest X-rays using hybrid CNN-transformer architecture leveraging LSTM feature selection},
  journal = {IET Conference Proceedings},
  year = {2026},
  volume = {2025},
  number = {63},
  pages = {146--150},
  doi = {10.1049/icp.2026.2352}
}
Kostenlos starten →