Abstract
Telehealth monitoring is an innovative process of synergising the benefits of information and communication technologies (ICT) and Internet of Things (IoT) to deliver healthcare services to remote, distant, and underserved regions. The objective of this study is to deliver healthcare services to patients outside the conventional settings by connecting the patient and healthcare providers with technology. As technologies for telehealth monitoring have become more advanced, they have become fully integrated into delivery of healthcare service. One of the most publicized telehealth services is the use of telemammography in the early diagnosis of breast cancer from remote and rural locations. Automated detection and classification of tumor in telemammographic images is of high importance for physicians for accurate prediction of the diseases. This study presents advances in telehealth services and also proposes novel telemammography system for early detection of breast cancer from remote and underserved areas. In this study, we have used efficient wavelet-based image processing techniques for preprocessing, detection, and enhancing the resolution of mammographic images. A detailed comparative analysis is performed to select the best classification model using different classification algorithms. We have used Multi-Layer Perceptron Neural Networks, J48 decision trees, Random Forest, and K-Nearest Neighbor classifier for classifying the tumor into three categories namely: benign, malignant, and normal. The classification is based on the area, volume, and boundaries of tumor masses. All the tumor features and classification methods are compared using Accuracy, Sensitivity, Specificity, Precision, and Mean Square Error. Experimental results on the Mammographic Image Analysis Society (MIAS) database are found to give the best results when neural network classifier is used for classification of mammographic images.