Classification in Thermograms for Breast Cancer Detection using Texture Features with Feature Selection Method and Ensemble Classifier

Abstract

the most common cancer among the women is breast cancer with very high mortality rate accounting for about 7% of the all cancer deaths (1). Though very nominal, the men too can have the chances of developing the breast cancer. The early detection can be boon for survival chances of the patients. Though Mammography is commonly accepted screening tool technique for breast cancer detection. But the thermography has the advantage of the early detection of the cancer when no masses are formed to be detected by the mammography. Moreover, mammography is a painful procedure and patient is exposedto harmful X-rays. The thermography is based on the asymmetry between affected and the normal breasts due to increased blood flow in the cancerous cells. This results in the difference in the temperature profile of the two breasts which is detected with the help of thermal imagers. The texture of bothbreasts are obtained withGabor texturefeatures. The features that can contribute to the classification are selected from the feature space of the all Gabor features extracted. Finally, the classification of the thermograms into healthy and sickcases are done using ensemble classifier. The accuracy obtained in his paper using selected Gabor features and ensemble classifier is 92.55%.