Optimized Deep Learning Model using Medical Imaging for CAD Applications
Year: 2025
Type: conference-paper
Abstract: Since the COVID-19 outbreak in 2019, its rapid global spread led to significant losses across various sectors and disrupted medical facilities worldwide due to the surge in cases. The shortage of doctors highlighted the need for IoMT applications to assist medical professionals in early and accurate decision-making. This paper presents an approach to developing compressed deep learning models with ensemble XAI (Explainable Artificial Intelligence) for SARS-CoV-2 detection using CXR and CT-Scan images. The proposed approach addresses the constraints of deep learning models, such as high computational complexity and lack of interpretability, while maintaining accuracy in identifying COVID-19 cases. Techniques like model compression, pruning, quantization, and ensemble XAI (including GradCAM++ and SHAP) are employed to improve robustness and interpretability. Our proposed optimized DenseNet-201 model achieved an accuracy of 97.07%, while reducing model size by 25%. Experimental results demonstrate that the method achieves competitive results while significantly reducing computational complexity.
Source: Procedia Computer Science
Authors: Deepanshi, Sameer Malik, Madhuri Gupta, Deepika Pantola
Institution: Bennett University
Language: English
Topic: COVID-19 diagnosis using AI
Subfield: Radiology, Nuclear Medicine and Imaging
Field: Medicine
Domain: Health Sciences
Open Access status: diamond
OpenAlex ID: W4410258103