An Extensive Review on Emerging Advancements in Thermography and Convolutional Neural Networks for Breast Cancer Detection

Breast cancer remains a significant health concern, necessitating early and accurate detection methods to reduce mortality rates. This review examines the use of thermography for breast cancer detection, highlighting the application of Convolutional Neural Networks (CNNs) to enhance diagnostic accuracy. Thermography, a non-invasive and cost-effective method, detects temperature variations using infrared radiation, demonstrating recall rates exceeding 90% and true negative rates over 90%. Advanced CNN models, such as DenseNet201 and ResNet101, achieved 100% accuracy in detecting breast cancer from thermal images. Techniques like Multi-Layer Perceptron Neural Network (MLP-NN), Support Vector Machine (SVM), and Artificial Neural Network (ANN), optimized through methods like weight-based ensemble feature selection and stochastic gradient descent, significantly improved detection accuracy. For example, the inception MV4 model reached an accuracy of 99.75% with a runtime of 7.7 min. These findings suggest that integrating CNNs with thermography provides a robust and efficient method for early breast cancer detection, which can be applied in clinical settings for routine screening and diagnosis.