Fusion of Deep Learning and Image Processing Techniques for Breast Cancer Diagnosis

Abstract

Deep learning has the capacity to gain great accuracy of diagnosing of numerous types of cancers, along with lung, cervical, colon, and breast cancer. It builds an efficient algorithm based on multiple processing (hidden) layers of neurons. Manual assessment of Cancer using Medical Image (CT images) requires expensive human labors and can easily cause the misdiagnose of any type of cancer. The Researcher focus on automatically diagnosing cancer by using the deep learning technique. Breast cancer is a particularly common sickness among women and maximum associated cause of female mortality. The survival rate of breast cancer patients can be expanded with the aid of powerful treatment, which can initiate upon early prognosis of the disease. This chapter introduces Deep Learning under medical image processing to analysis and diagnosis of Cancer (Ehteshami Bejnordi et al., in Deep learning-based assessment of tumor-associated stroma for diagnosing breast cancer in histopathology images, pp. 929–932, 2017 [1]). Identification of most cancer might facilitate in sparing a massive wide variety of lives over the globe community and deep neural networks may be correctly used for intelligent image analysis. The essential structure of how this deep learning takes a shot at medical image processing (Litjens et al. in A survey on deep learning in medical image analysis, 2017, [2]; Rezaeilouyeh et al. in J Med Imaging 3(4):044501, 2016 [3]) is furnished in this research, i.e., pre-processing, image segmentation and post-processing. The following piece of this part depicts the rudiments of the field of disease conclusion, which incorporates steps of malignant growth determination followed by the regular arrangement strategies utilized by specialists, giving a verifiable thought of disease grouping methods to the readers. Next an attempt has been made to classify the extracted features from mammograms as benign or malignant by using Convolutional neural network (CNN) (Ciresan in Mitosis detection in breast cancer histology images with deep neural networks. Springer, Berlin, pp. 411–418, 2013 [4]; LeCun et al. in International symposium on circuits and systems, pp. 253–256, 2010 [5]; Huynh et al. in J Med Imaging 3(3):034501, 2016 [6]) is applied to classify cancer using optimal features obtained from cell segmented images. Performance improvised of the approaches by varying various parameters is studied.