Breast Cancer Detection Through Gabor Filter Based Texture Features Using Thermograms Images

The mortality rates in women is highest due to breast cancer among other all the cancers in developed as well as in developing countries. As evident from the facts that mortality rate of 12.7 among 1, 00, 000 in India [1] and whereas in USA, estimated deaths of 40,610 women i.e. 6.8% of all cancer deaths in 2017. Mammography is considered to be the most accepted technique for breast cancer detection. In this paper, thermography is explored as a viable alternative to the mammography. As mammography has its own drawbacks of being a painful procedure, exposure of the body to harmful Xrays. This necessitates in exploring the other modalities preferably non-contact and without using any harmful radiations. Thermography is coming out to be an alternative to the standard mammography with advantages of being noninvasive, safe, portability and cost effectiveness. The temperature pattern of the breasts changes as a result of the high increased blood flow into affected cells. This gives the way to asymmetry between normal and cancerous breast which can be detected using different techniques. In this paper, 35 normal and 35 abnormal thermograms are taken from on line DMRDatabase for Mastology Research having breast thermograms for early detection of breast cancer. The texture features of the left and right breasts are extracted using Gabor filters. The thermograms are then classified using support vector machine (SVM) based on the textural asymmetry between the breasts into normal and cancerous cases. The accuracy achieved using Gabor features and SVM classifier is 84.5% The early detection of cancer using thermography increases the survival chances of the patient considerably as it can detect the cancer in initial stages.