Detection and Classification of Breast Lesions in Biomedical Images in an Explainable Way

Purpose: Breast cancer is a major public health issue and the leading cause of cancer death among adult women in Brazil. Late diagnosis contributes to high mortality rates and invasive surgeries. Limited access to imaging tests results in diagnoses at advanced stages, underscoring the need for early detection to improve cure rates and avoid aggressive procedures. The objective of this study is to enhance breast cancer diagnosis, particularly in countries with prevalent late diagnoses, by applying advanced artificial intelligence and extreme learning techniques to thermographic images. Methods: The study utilizes advanced artificial intelligence (AI) and extreme learning techniques to analyze thermographic images. The goal is to develop a computer system that not only accurately diagnoses breast cancer but also provides explanations for its decisions in a way that healthcare professionals can understand. Results: The proposed AI approach demonstrated efficiency in detecting and classifying breast lesions in thermographic images. The system achieved an accuracy of 89.70% in distinguishing malignant lesions from other diagnoses in its best-case scenario. Conclusion: The developed system represents a significant advancement in breast cancer diagnosis, particularly in regions where late-stage diagnoses are common. It holds potential to improve diagnostic accuracy, contribute to better patient outcomes, and enhance recovery prospects for breast cancer patients.