Automated COPD Diagnosis from CT Scans: A Hybrid Deep Learning and Machine Learning Approach with Explainable AI

A. Hema, C. H. Hussaian Basha, S. Senthilkumar, B. S. Gopika, R. Muthaiyan, R. Ramanan

2025

Abstract

Chronic obstructive pulmonary disease (COPD) is a widespread and debilitating disease of the lungs that requires the patient to endure, requiring a precise and timely diagnosis to aid in care. In this research, an innovative method is introduced that combines the classical machine learning methods with feature extraction based on deep learning to detect and classify the COPD severity from Computed tomography (CT) scans. We employ CNNs pretrained on large amounts of medical data to extract deep features showing indicative structural changes in the lung, including emphysema and thickening of airway walls and other morphological deformations. These features are then used by Support Vector Machine (SVM) to get the accurate COPD severity classification. This study uses Gradient Weighted Class Activation Mapping (Grad-CAM) and Shapley Additive explanations (SHAP) to explain the method prediction with the purpose of increasing the transparency and interpretability and increasing confidence in AI-driven diagnostics. The proposed methodology is validated on the LIDC-IDRI (Lung Image Database Consortium Image collection) dataset for emphysema severity and airway abnormalities. Results of the comparison show that this hybrid method is more accurate or robust than CNN or traditional ML methods alone. Results show the importance of explainable and efficient AI in medical imaging in early COPD detection, monitoring of drug efficacy, and severity assessment.

Paper Citation

in Harvard Style

Hema A., Basha C., Senthilkumar S., Gopika B., Muthaiyan R. and Ramanan R. (2025). Automated COPD Diagnosis from CT Scans: A Hybrid Deep Learning and Machine Learning Approach with Explainable AI. In Proceedings of the 1st International Conference on Research and Development in Information, Communication, and Computing Technologies - ICRDICCT`25; ISBN 978-989-758-777-1, SciTePress, pages 208-217. DOI: 10.5220/0013910500004919

in Bibtex Style

@conference{icrdicct`2525,
author={A. Hema and C. Basha and S. Senthilkumar and B. Gopika and R. Muthaiyan and R. Ramanan},
title={Automated COPD Diagnosis from CT Scans: A Hybrid Deep Learning and Machine Learning Approach with Explainable AI},
booktitle={Proceedings of the 1st International Conference on Research and Development in Information, Communication, and Computing Technologies - ICRDICCT`25},
year={2025},
pages={208-217},
publisher={SciTePress},
organization={INSTICC},
doi={10.5220/0013910500004919},
isbn={978-989-758-777-1},
}

in EndNote Style

TY - CONF

JO - Proceedings of the 1st International Conference on Research and Development in Information, Communication, and Computing Technologies - ICRDICCT`25
TI - Automated COPD Diagnosis from CT Scans: A Hybrid Deep Learning and Machine Learning Approach with Explainable AI
SN - 978-989-758-777-1
AU - Hema A.
AU - Basha C.
AU - Senthilkumar S.
AU - Gopika B.
AU - Muthaiyan R.
AU - Ramanan R.
PY - 2025
SP - 208
EP - 217
DO - 10.5220/0013910500004919
PB - SciTePress