An Efficient Approach for Diagnosing the Breast Cancer Using Deep Learning Technique

Abstract

Background: Now a days breast cancer is one of the deadliest diseases found in most of the matured women. The Cancer disease is curable if it is diagnosed at initial stages. But once it goes to the final stages it is very hard to cure and which leads the patient to death. 

Objectives:In this paper, we proposed a model to diagnosis the cancer which gives a clinical support to the physician for initial diagnosis of breast cancer. In general, the cancer disease is identified if there is a tumour growth is appeared in the human body. But before the tumour grows, there may be change in textures of the different biological parts in the region of breast where tumours can be grown. Here, in this paper, we propose an image processing technique to detect the change in structural parameters of the mammography images.

Methods:The proposed image processing technique and convolutional neural network were combined here which forms a layered approach in deep learning.  Here, we prefer manual prediction of the images rather than automatic bulky predictions so as to ensure an image is correctly predicted whether it is malignant or benignant.

Results and Conclusion: As based on the experimental results, we prove that the proposed work attains outstanding results compared to the ideal CNN approach in terms of wide variety of parameters such as accuracy of detection, Sensitivity, Specificity, Positive Predictive Value (PPV), Negative Prediction Value (NPV), Mathew’s Correlation Coefficient(MCC).