Ensemble Transfer Learning Approach for Breast Cancer Classification using Thermograms

Breast cancer is one of the primary probable sicknesses particularly among women and utmost associated agent of female mortality. The survivability of breast cancer patients is increased since the last decade with the help of powerful treatment, which is made possible due to advancement in technology. Imaging is one of the important parts of cancer clinical protocol. Thermal imaging is a non-invasive, non-contact and economical imaging technique showing promising results. Transfer learning reduces the computational time required in training while building a network or model for image classification. This paper proposes a methodology using ensemble transfer learning approach for breast thermogram classification as cancerous and non-cancerous. Competitive results were obtained and compared with existing works. Using the proposed approach, a maximum accuracy of 96.25% and F1 score of 0.96 were obtained.