AI‐Driven Healthcare Analysis
N Kasthuri, T. Meeradevi · 2022
AI plays a major role in almost all fields. Healthcare is one of the fields where AI can outperforms the existing techniques used in healthcare technologies. Breast cancer identification at the earliest stage is a challenging task where AI can be applied to determine whether the cancer cells are present or not with highest accuracy. As per 2013, World Health Organization survey statistics reported that approximately 508,000 women died due to breast cancer across the world during the year 2011. It is possible to identify the breast cancer in its primary stage, and hence, it can be healed. But, the cancers are identified at last stage for most of the women. Early recognition of breast cancer is very important to increase the survival rate. Machine learning technique and technologies are used for the identification and classification of cancer in an accurate level. Classifier such as stochastic gradient decent (SGD), support vector machine (SVM), nearest neighbors (KNN), naïve Bayes, random forest, and convolutional neural network (CNN) are the various machine learning techniques that can be used for the classification of various tumors. The dataset has been taken from Mammography MIAS database. There are totally 322 images. The Gray Level Co-Occurrence Matrix (GLCM) are used to extract the features from the images, thereby converting the useful information from image space to feature space. These features are used to classify the presence or absence of cancer cells using Machine learning classifiers. Two different classes of breast cancer tumors are categorized as benign and malignant. Invasive ductal carcinoma and invasive lobular carcinoma are considered as the main category of breast cancer. The breast cancer which is curable is ductal carcinoma when it is found at earliest stage. The other type that can rapidly spread to lymph nodes and other parts of the body and starts in a lobule of the breast is called as invasive lobular carcinoma. In addition to mammogram images, the histopathology images are taken for analysis and the deep learning models are used to provide better classification. The images are fed as an input to these models and the features are extracted using the principles of convolution and these features are used to classify the images. The Kaggle dataset is used to train the deep learning model. Breast histopathology images around 200,000 images are available; hence, it is possible to train the CNN classifier with higher accuracy to classify Benign (non-cancerous) and malignant (cancerous).