EUVIMEDEUVIMED entdecken
Crossref

A Comparative Evaluation of Convolutional Neural Network Models for Automated Tuberculosis Detection Using Chest Radiographs

Ruchika Nagar, Dr. Rajendra N Solanki, Dr. Dipak Patel, Dr. Vibha Patel, Dr. Manish Patel

International Journal of Drug Delivery Technology · 12.05.2026 · 16 (26s)

DOI: 10.25258/ijddt.16.26s.32 ↗

Kurz gefasst

Fachartikel in der Zeitschrift International Journal of Drug Delivery Technology, online erschienen am 12.05.2026. Verfasst von Ruchika Nagar und 4 weiteren Autor:innen.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

This paper compares the performance of six state-of-the-art CNN architectures, fine tuned to detect Tuberculosis (TB) from chest X-ray images. We used a combined dataset of 3704 images from four different public sources (Montgomery, Kaggle, Shenzhen, TBX11K) and used these to train six state-ofthe-art CNN models in this study—Resnet50, Densenet121, EffcientnetV2, Xception, InceptionV3 and Mobilenetv2— to distinguish between TB and Normal cases. These six models were tested using accuracy, sensitivity, specificity, precision, and area under the curve (AUC) on a private hospital’s test data which contained 87 images that have never been seen before. Each model demonstrated divergent training loss and validation loss, thus illustrating how the limitations of each model generalize on the small dataset we had available. Using the same metrics as our test data (Accuracy, Sensitivity, Precision, F1 score, AUC) the Xception model was shown to be the best performer at 95.45% F1, 96.56% AUC, and 42 seconds of inference time; whereas, the newest EfficientNetV2 model produced a much lower AUC of 78.80%.The three best performing architectures of this study were Xception, InceptionV3, and Resnet50, based upon their performance on the private hospital's test data. The Xception model produced the best results for both sensitivity and precision, which is confirmed by the use of grad-cam for visualizing the pulmonary opacities.

Zitieren

APA

Nagar, R., Solanki, D. R. N., Patel, D. D., Patel, D. V., & Patel, D. M. (2026). A Comparative Evaluation of Convolutional Neural Network Models for Automated Tuberculosis Detection Using Chest Radiographs. International Journal of Drug Delivery Technology, 16(26s). https://doi.org/10.25258/ijddt.16.26s.32

Vancouver

Nagar R, Solanki DRN, Patel DD, Patel DV, Patel DM. A Comparative Evaluation of Convolutional Neural Network Models for Automated Tuberculosis Detection Using Chest Radiographs. International Journal of Drug Delivery Technology. 2026;16(26s). doi:10.25258/ijddt.16.26s.32

BibTeX
@article{Nagar2026,
  author = {Nagar, Ruchika and Solanki, Dr. Rajendra N and Patel, Dr. Dipak and Patel, Dr. Vibha and Patel, Dr. Manish},
  title = {A Comparative Evaluation of Convolutional Neural Network Models for Automated Tuberculosis Detection Using Chest Radiographs},
  journal = {International Journal of Drug Delivery Technology},
  year = {2026},
  volume = {16},
  number = {26s},
  doi = {10.25258/ijddt.16.26s.32}
}
Kostenlos starten →