Evaluating color space in thermography for enhanced breast cancer detection using deep learning

Al Husaini, Mohammed Abdulla Salim and Habaebi, Mohamed Hadi and Al Husaini, Yousuf Nasser and Abdulghafor, Rawad and Abrar, Mohammad (2025) Evaluating color space in thermography for enhanced breast cancer detection using deep learning. In: 2025 IEEE 10th International Conference on Smart Instrumentation, Measurement and Applications (ICSIMA), KL Malaysia.

PDF - Published Version Restricted to Repository staff only Download (873kB) | Request a copy

Abstract

Breast cancer is a leading cause of cancer-related mortality among women worldwide, with early detection playing a critical role in improving survival rates. Traditional screening methods such as mammography face limitations, including false positives, false negatives, and restricted accessibility, especially in low-resource settings. This study evaluates the potential of thermography combined with deep learning for enhanced breast cancer detection, focusing on the impact of different color spaces. A dataset comprising thermographic images of 287 individuals, captured using a FLIR SC 620 thermal camera, was transformed into various color spaces, including RGB, HSV, LAB, and CMYK, among others. Inception MV4, a convolutional neural network (CNN), was employed to assess the effectiveness of these color spaces in detecting breast cancer. The CMYK color space demonstrated superior performance with a validation accuracy of 98.18%, followed by LAB (95.07%). In contrast, the grayscale color space yielded the lowest accuracy (87.48%). Results indicate that color spaces significantly influence model performance, with CMYK emerging as the most efficient in terms of accuracy and processing time. The study concludes that utilizing alternative color spaces in thermography can substantially improve the accuracy of breast cancer detection using deep learning. Future work will focus on optimizing the model's performance by exploring additional color spaces, reducing computational complexity, and validating the approach on larger, more diverse datasets.

Item Type:	Proceeding Paper (Plenary Papers)
Uncontrolled Keywords:	Breast Cancer, Thermal Image, Color Space, Deep Learning.
Subjects:	T Technology > TK Electrical engineering. Electronics Nuclear engineering
Kulliyyahs/Centres/Divisions/Institutes (Can select more than one option. Press CONTROL button):	Kulliyyah of Engineering > Department of Electrical and Computer Engineering Kulliyyah of Engineering
Depositing User:	Dr. Mohamed Hadi Habaebi
Date Deposited:	21 Nov 2025 15:14
Last Modified:	21 Nov 2025 15:14
Queue Number:	2025-11-Q068
URI:	http://irep.iium.edu.my/id/eprint/124492

Actions (login required)

View Item