DeepThorax: Multi-Disease Chest X-ray Detection and Clinical Decision Support System | IJCT Volume 13 – Issue 5 | IJCT-V13I5P58
International Journal of Computer Techniques
ISSN 2394-2231
Author
Mrs. Bindu M, Gagan D S, Rahul M S, Safwan Mansoor, R Jeevan Kumar
Abstract
Chest X-ray analysis is widely used to examine thoracic abnormalities, but reviewing large numbers of images can be time-consuming and interpretations may differ between readers. This paper presents Deep Thorax, an AI-assisted prototype for multi-disease chest radiograph analysis. It uses an ImageNet-pretrained DenseNet121 model to classify fourteen thoracic findings in the NIH ChestX-ray14 dataset. A sigmoid output is assigned to each finding, allowing more than one condition to be predicted for an image. The system also uses Grad-CAM to visualize image regions influencing predictions and displays patient context, including age, gender, smoking history, symptoms, and medical history. Additional functions provide nearby healthcare-facility suggestions, generate a text report, and retain previous analyses for reference. The ROC curve in this study reports ROC-AUC values of 0.998 for COVID-19, 0.966 for Normal, 0.973 for Pneumonia, and 0.999 for Tuberculosis. These results suggest strong discrimination on the evaluated split, but external validation and radiologist assessment are still required before clinical use. DeepThorax is intended to support professional review and is not a replacement for radiologists or a definitive diagnostic system. The current prototype also separates patient-supplied context from the image classifier, so these details are provided for review rather than claimed as factors that improve prediction accuracy.
Keywords
chest X-ray classification; DenseNet121; multi-label learning; Grad-CAM; explainable AI; clinical decision support; NIH ChestX-ray14
Conclusion
This paper presented DeepThorax, a prototype pipeline that combines DenseNet121-based multi-label classification of fourteen ChestX-ray14 thoracic findings with Grad-CAM explainability, contextual patient information, hospital recommendation, AI-assisted reporting and patient history tracking. Preliminary ROC-AUC results range from near-chance to moderate discrimination across the fourteen findings, indicating that the classifier requires further training, tuning and validation before it could support any clinical use. The main contribution of this work is the integration of classification, explainability and clinical-workflow features into a single accessible system rather than a claim of state-of-the-art detection accuracy. DeepThorax is presented strictly as an engineering decision-support prototype and not as a deployment-ready or diagnostic tool.
References
[1]M. J. P. Priyadarsini et al., “Lung Diseases Detection Using Various Deep Learning Algorithms,” Journal of Healthcare Engineering, vol. 2023, Article ID 3563696, 2023.
[2]S. K. H. Bukhari and L. Fahad, “Lung Disease Detection Using Deep Learning,” in Proceedings of the International Conference on Emerging Technologies (ICET), 2022.
[3]J. Zhu et al., “Multiple Lung Diseases Detection Using Advanced Deep Learning Model with Attention Mechanisms and Upsampling Features,” Engineering Applications of Artificial Intelligence, vol. 156, 2025.
[4]M. Patel and M. Shah, “Deep Ensemble Learning Model for Diagnosis of Lung Diseases from Chest X-Ray Images,” Indian Journal of Science and Technology, vol. 17, no. 8, pp. 702–712, 2024.
[5]A. Nadal-Martínez et al., “Lung Disease Classification Using Deep Learning and ROI-Based Chest X-Ray Images,” Technologies, vol. 14, no. 1, 2026.
[6]P. S. Kulkarni et al., “Lung Disease Detection Using Deep Learning Algorithms,” Biomedical Engineering Open Access, vol. 1, no. 1, 2025.
[7]P. Kwaśniewska et al., “Pulmonary Diseases Identification: Deep Learning Models and Ensemble Learning,” Applied Computer Science, vol. 21, no. 3, pp. 38–58, 2025.
[8]M. Irtaza et al., “Multi-Label Classification of Lung Diseases Using Deep Learning,” IEEE Access, vol. 12, 2024.
[9]S. N. Pranavi and V. Sreedevi, “Lung Disease Detection Using Deep Learning,” Bachelor of Engineering project report, Sathyabama Institute of Science and Technology, 2023.
Y. S. Kumari, “Deep Learning Approaches for Lung Disease Detection Through Voice Analysis,” International Journal of Creative Research Thoughts (IJCRT), vol. 12, no. 10, 2024
How to Cite This Paper
Mrs. Bindu M, Gagan D S, Rahul M S, Safwan Mansoor, R Jeevan Kumar (2026). DeepThorax: Multi-Disease Chest X-ray Detection and Clinical Decision Support System. International Journal of Computer Techniques, 13(5). ISSN: 2394-2231.
