Personalized medical image indexing with textual descriptions using SVM based unsupervised classification framework

摘要

Electronic medical record (EMR) system usually works as an application for collecting patients' diagnostic information and support medical decision making processing. Many EMR systems have personalized indexing support to provide concept specific searching facility. The medical images provide greater evidence about the nature and harm of the deceases. For analyzing the content of the medical images, the traditional content based feature extraction and retrieval mechanism with an active learning process can be used. We proposed a simple and highly reliable framework for extracting medical image features, then learning the association between the features and descriptions provided by the physicians. The integrated c-SVM (classifier -Support Vector Machine) based classifier is used to build a framework that provides a decision model for classification and decision-making. And this framework can also be used to classify new images and to build the personalized index for retrieving medical images for a desired description.