Intelligent Computing Using Deep Learning for Screening of Breast Cancer from Breast Thermograms

Breast cancer is the second leading cause of deaths among women due to cancer. The WHO reports that 2.3 million women have been diagnosed with breast cancer, and that this has resulted in 685,000 deaths worldwide, with a mortality rate of 19.9 per 100,000 women each year (Estebsari et al. Protection motivation theory and prevention of breast cancer: a systematic review, [1]). Mammography is a painful technique of determining breast cancer. It also has its X-rays effect on the breast. Thermography turned out to be a gameplayer in deciding breast cancer. Thermal radiation coming out of breast is studied to determine the cancer possibility. It is painless and inexpensive. Survival rate is increased if this disease is diagnosed at an early stage. Breast cancer can be cured with 97% chances of survival if it is detected earlier (Madhavi and Thomas in Quant Infrared Thermogr J 16(1):111–128, 2019, [2]). VGG-16 (Simonyan and Zisserman in Very deep convolutional networks for large-scale image recognition, 2014) and ResNet-101 (Xu et al. in 2020 IEEE 2nd international conference on civil aviation safety and information technology, ICCASIT, Weihai, 2020, [3]) neural network is trained with breast thermogram image dataset which is used to classify new thermograms as sick and healthy thermograms. Accuracy of 100, 98.31, 98.31% is obtained with ResNet101 with SGDM, ADAM, RMSprop optimizers. Accuracy of 98.31% is obtained with VGG 16 while using SGDM, ADAM, and RMSprop optimizers. The cancerous region is also detected from the thermogram using MATLAB coding.