Evaluating the effect of noisy thermal images on the detection of early breast cancer using deep learning

Al-Hussaini, Mohammed Abdulla Salim and Habaebi, Mohamed Hadi and Islam, Md. Rafiqul and Al Husaini, Yousuf Nasser (2025) Evaluating the effect of noisy thermal images on the detection of early breast cancer using deep learning. Advances in Artificial Intelligence and Machine Learning, 5 (2). pp. 3923-3953. ISSN 2582-9793

PDF - Published Version Restricted to Registered users only Download (12MB) | Request a copy

Abstract

Breast cancer remains a leading cause of mortality among women globally. Some techniques have been developed to enhance early detection, among which thermal imaging has emerged as a promising modality capable of identifying potential signs of breast cancer in its early stages. In addition, Thermal images provide valuable pixel-level information by capturing temperature variations between healthy and cancerous tissues. However, the susceptibility of these thermal images to noise poses a challenge to diagnostic accuracy in the early stages. This research aims to assess the influence of various types of noise on the performance of recently developed deep-learning models designed for early breast cancer detection. In addition, a comprehensive analysis was conducted using a substantial dataset to assess the impact of noise on the models’ efficacy. It also encompasses different categories of noise, which are characterised by distinct mean and variance values ranging from 0.01 to 0.09. The findings reveal that introducing various types of noise, albeit within a small range of mean and variance values, adversely affects the performance of deep learning models. It shows that these filters play a pivotal role in enhancing classification accuracy. Moreover, the results show that salt and pepper noise varied between 0.1 and 0.3, significantly impacting the accuracy of Inception MV4, reducing it from 100% to 51.58%, without adding filters in pre-processing. Additionally, introducing variance in multiplicative noise from 0.2 to 0.8 demonstrated an effect on classification accuracy only at noise levels of 0.7 (89%) and 0.8 (43%). Moreover, the results show that performance metrics for the proposed method were accuracy of 99.975%, sensitivity of 0.994, specificity of 1, precision of 1, NPV of 0.995, FNR of 0.006, LRN of 0.006, AUC of 0.997, EER of 0.003, and F1 score of 0.997, but FPR of 0. In conclusion, findings underscore the significance of refining noise mitigation strategies and pre-processing techniques to advance the reliability and accuracy of thermal imaging as a diagnostic tool in breast cancer detection in the early stages.

Item Type:	Article (Journal)
Uncontrolled Keywords:	Breast cancer, thermal image, gaussian noise, deep learning
Subjects:	T Technology > TK Electrical engineering. Electronics Nuclear engineering > TK7800 Electronics. Computer engineering. Computer hardware. Photoelectronic devices
Kulliyyahs/Centres/Divisions/Institutes (Can select more than one option. Press CONTROL button):	Kulliyyah of Engineering > Department of Electrical and Computer Engineering
Depositing User:	Dr. Mohamed Hadi Habaebi
Date Deposited:	27 Jun 2025 22:03
Last Modified:	27 Jun 2025 22:03
URI:	http://irep.iium.edu.my/id/eprint/121738

Actions (login required)

View Item