Machine Learning approach for Numerical technique for classifying Cysts and Malignant Tumors in Breast

Abstract Tumor is an abnormal tissue which can be appeared at any part of the body. It can be classified to either benign or malignant. One of the most common women's tumors that infest the breast. Various benign disorders like development of cysts in woman’s breast occur due to hormonal changes and are at the risk of becoming malignant. A number of thermal models are reported to differentiate between normal and malignant tissues of breast. But no thermal model is reported in study the effect of benign disorders on the literature to distinguish between benign and malignant disorders in woman’s breast. An attempt has been made in this paper to study the thermal disturbances caused by cysts and malignant tumors in the fat tissues of woman’s breast. The model is developed for a two-dimensional steady state case using penne’s bio heat equation and incorporating parameters like thermal conductivity, blood mass flow rate and self-controlled metabolic heat generation. The appropriate adiabatic boundary conditions have been framed for various environmental conditions. The finite element method has been employed to obtain the solution. The results have been obtained for different sizes of spherical shaped cysts and different depth of tissues in hemispherical shaped woman’s breast. The relation of size and position of the cysts have been studied with the thermal distribution in various tissues layers of the woman’s breast. The comparison of thermal profiles for cysts and malignant tumors in woman’s breast has been performed. A contrast in thermal behavior of cyst and malignant tumor in woman’s breast is observed which can be useful to distinguish between the malignant tumor and cyst in woman’s breast to prevent false positive test for malignant tumor. Accordingly, this study found that there are various factors that could affect the cancer classification and prediction. Therefore in this study, Breast cancer data classification have been done using three classification techniques which are Artifical Neural Network (ANN), Support Vector Machine (SVM), and Random Forest (RF) in order to improve the performance of the model trained the model with selected features according to the analysis done.