Citation

Abdullahi, Muhammad Kabir and Mansor, Sarina and Fauzi, Mohammad Faizal Ahmad and Wadood, Arbab Sufyan and Nabi, Md Serajun (2026) Enhancing Nasopharyngeal Carcinoma Diagnosis: A Comparative Study of Transfer Learning and Non-Transfer Learning on Whole Slide Image Datasets. In: 2026 IEEE 16th Symposium on Computer Applications & Industrial Electronics (ISCAIE), 25-26 April 2026, Penang, Malaysia.

Text 8.pdf - Published Version Restricted to Repository staff only Download (477kB)

Abstract

Nasopharyngeal carcinoma (NPC) remains a challenging malignancy to diagnose early due to its subtle symptoms, and complex tissue structures. This study aims to enhance automated NPC detection through a comparative evaluation of eight state-of-the-art convolutional neural network (CNN) architectures, DenseNet201, VGG16, ResNet50, MobileNet, Xception, InceptionV3, EfficientNetB0, and NASNetMobile, using both transfer learning (TL) and non-transfer learning (NTL) approaches. Experiments were conducted on two datasets: a private whole-slide image (WSI) dataset collected from Sarawak General Hospital and Hospital Kuala Lumpur, and a public Breast Cancer Imaging (BCI) dataset. The methodology involved systematic training and evaluation under four configurations: Transfer Learning and Non-Transfer Learning, each with and without data augmentation, to assess the influence of training strategy and data diversity. The results indicate that model performance varies significantly across architectures and datasets, with DenseNet201 and EfficientNetB0 consistently demonstrating strong generalization. Overall, the study highlights the importance of model-dataset alignment and the combined role of transfer learning and augmentation in improving diagnostic performance for histopathological NPC classification, thereby directly supporting downstream analysis in the Nasopharyngeal Carcinoma context.

Item Type:	Conference or Workshop Item (Paper)
Uncontrolled Keywords:	Nasopharyngeal carcinoma, Lymphoid hyper plasia, Nasopharyngeal inflammation, CNN, Transfer learning, Whole slide images, Deep learning, Histopathology
Subjects:	Q Science > QA Mathematics > QA71-90 Instruments and machines
Divisions:	Faculty of Artificial Intelligence & Engineering (FAIE)
Depositing User:	Ms Suzilawati Abu Samah
Date Deposited:	04 Aug 2026 07:29
Last Modified:	04 Aug 2026 07:29
URII:	http://shdl.mmu.edu.my/id/eprint/16507

Downloads

Edit (login required)