Enhancing Histopathological Breast Cancer Image Classification using Deep Learning
Puspanjali Mohapatra1, Baldev Panda2, Samikshya Swain3
1Prof. Puspanjali Mohapatra, Department of Computer Science and Engineering, International Institute of Information Technology, Bhubaneswar (Odisha), India.
2Baldev Panda, Department of Electronics & Telecommunication, International Institute of Information Technology, Bhubaneswar (Odisha), India.
3Samikshya Swain, Department of Electronics & Telecommunication, International Institute of Information Technology, Bhubaneswar (Odisha), India.
Manuscript received on 01 May 2019 | Revised Manuscript received on 15 May 2019 | Manuscript published on 30 May 2019 | PP: 2024-2032 | Volume-8 Issue-7, May 2019 | Retrieval Number: G6270058719/19©BEIESP
Open Access | Ethics and Policies | Cite | Mendeley | Indexing and Abstracting
© The Authors. Blue Eyes Intelligence Engineering and Sciences Publication (BEIESP). This is an open access article under the CC-BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/)
Abstract: In this paper we have conducted experimental analysis in predicting IDC (Invasive Ductal Carcinoma) as well as well as Malignant and Benign tumors from textual and histopathology image datasets. The analysis commences with the conventional machine learning algorithms on the text dataset and upgrades to deep learning while dealing with histopathology images. The machine learning algorithms like Logistic Regression, SVM, KNN, Decision tree are applied on the datasets to compare the accuracy among them. The model giving the best accuracy is decided through Feature extraction techniques like PCA and LDA leading to an improvement in accuracy. When dealing with large datasets consisting of high-resolution images, the machine learning algorithms don’t perform well. Deep learning has the ability to handle such complex situations which include high-dimensional matrix multiplications. Various architectures of CNN were applied and the model with the high generalization accuracy and minimal complexity is selected. The histopathology images are given as input to the CNN network as training models and then finally classified as having IDC or Malignancy. The best model is selected after varying the number of hidden layers and then applied to the dataset for final classification
Keyword: Breast cancer, IDC, Histopathology Images, Machine Learning, Logistic Regression, SVM, KNN, Random Forest, Deep Learning, CNN.
Scope of the Article: Deep Learning.