A CNN-Based Approach for Early Detection of Breast Cancer Using Infrared Imaging

Breast cancer remains a disease that affects women’s health worldwide and is a major cause of both new cancer cases and cancer-related deaths. Any advancement in early detection is crucial for both improved survival and the effectiveness of the appropriate treatments. This research aims to investigate the use of CNNs in infrared image-based breast cancer diagnosis, as well as the application of deep learning to improve cancer diagnosis outcomes. The study uses a number of related databases, mammography pictures, and histopathological data, including information from The Cancer Genome Atlas (TCGA) and the Digital Database for Screening Mammography (DDSM). Among the techniques frequently employed in the field of CNNs to enhance the model’s performance and work with the input information are normalisation and augmentation. Convolutional, pooling, and fully connected layers are the main layers used by the CNN structure to extract characteristics from and classify breast cancer images. The model’s specificity, which is assessed, is ${9 1 \%}$, and its recall and precision reveal a decent foundation that indicates the model can distinguish between benign and malignant instances. In addition, real-time thermal imaging sensor data and patient data acquired via RFID tags are used in the research to supplement the dataset and improve clinical applicability.