Citation

Mahamud, Eram and Assaduzzaman, Md. and Fahad, Nafiz and Hossen, Md. Jakir (2026) Deep-lymph: An advanced deep learning framework for precision diagnosis of lymphoma from histopathological images. Intelligence-Based Medicine, 13. p. 100347. ISSN 26665212

Text main.pdf - Published Version Restricted to Repository staff only Download (18MB)

Abstract

Deep learning models have shown great promise in medical image classification, but their lack of interpretability limits their adoption in clinical settings. This study addresses the need for explainable models in lymphoma diagnosis using an enhanced model within a transfer learning framework. To improve interpretability, we incorporated Explainable Artificial Intelligence (XAI) techniques, including SHAP, LIME, Grad-CAM, and GradCAM++, Occlusion Sensitivity Map to provide insights into the model's decision-making process. The model was trained on high-quality lymphoma imaging data, and preprocessing techniques such as Denoising, Contrast Limited Adaptive Histogram Equalization (CLAHE), and Gamma Correction were applied to improve image clarity. We conducted an ablation study to identify optimal parameters for the model. Our proposed model achieved an accuracy of 99.99%, with precision and recall rates of 100%, demonstrating its exceptional performance. SHAP and LIME helped in understanding the model's decisions, while Grad-CAM and Grad-CAM++ identified the crucial image features that influenced classification, enhancing transparency and trust in AIassisted lymphoma diagnosis. This study contributes to advancing the use of deep learning in oncology, offering a reliable and interpretable tool for lymphoma detection.

Item Type:	Article
Uncontrolled Keywords:	Deep learning, cancer diagnosis
Subjects:	R Medicine > R Medicine (General) > R858-859.7 Computer applications to medicine. Medical informatics
Divisions:	Faculty of Engineering and Technology (FET)
Depositing User:	Ms Rosnani Abd Wahab
Date Deposited:	02 Apr 2026 03:12
Last Modified:	02 Apr 2026 04:11
URII:	http://shdl.mmu.edu.my/id/eprint/15631

Downloads

Edit (login required)