Citation

Jayaram, Jayapradha and Haw, Su Cheng and Palanichamy, Naveen and Ng, Kok Why and Aneja, Muskan and Taiyab, Ammar (2025) EMI-LTI: An enhanced integrated model for lung tumor identification using Gabor filter and ROI. MethodsX, 14. p. 103247. ISSN 22150161

Text 1.pdf - Published Version Restricted to Repository staff only Download (4MB)

Abstract

In this work, the CT scans images of lung cancer patients are analysed to diagnose the disease at its early stage. The images are pre-processed using a series of steps such as the Gabor filter, contours to label the region of interest (ROI), increasing the sharpening and cropping of the image. Data augmentation is employed on the pre-processed images using two proposed architectures, namely (1) Convolutional Neural Network (CNN) and (2) Enhanced Integrated model for Lung Tumor Identification (EIM-LTI). • In this study, comparisons are made on non-pre-processed data, Haar and Gabor filters in CNN and the EIM-LTI models. The performance of the CNN and EIM-LTI models is evaluated through metrics such as precision, sensitivity, F1-score, specificity, training and validation accuracy. • The EIM-LTI model’s training accuracy is 2.67 % higher than CNN, while its validation accuracy is 2.7 % higher. Additionally, the EIM-LTI model’s validation loss is 0.0333 higher than CNN’s. • In this study, a comparative analysis of model accuracies for lung cancer detection is performed. Cross-validation with 5 folds achieves an accuracy of 98.27 %, and the model was evaluated on unseen data and resulted in 92 % accuracy.

Item Type:	Article
Uncontrolled Keywords:	Lung cancer, convolutional neural network
Subjects:	Q Science > Q Science (General) > Q300-390 Cybernetics R Medicine > RA Public aspects of medicine > RA421-790.95 Public health. Hygiene. Preventive medicine
Divisions:	Faculty of Computing and Informatics (FCI)
Depositing User:	Ms Rosnani Abd Wahab
Date Deposited:	28 Mar 2025 03:34
Last Modified:	28 Mar 2025 03:34
URII:	http://shdl.mmu.edu.my/id/eprint/13643

Downloads

Edit (login required)