Study and Definition of the Bag-of-Features Approach for Retrieval by Content of Medical Images Complying with the Specialists Expectations

Abstract

This project aims at studying and exploring the Bag-of-Features (BOF) approach to support content-based image retrieval for medical systems, considering the specialists' point of view and knowledge. Therefore, we intend to design new feature extraction methods from the images, and new ways of categorizing the features and the images, in order to build a more robust and efficient mechanisms to process similarity queries that comply with the specialists' interests. We will design a theoretical model to guide the development of perceptual CBIR systems, and propose more efficient ways of indexing images employing the BOF approach, aiming at improving the usability and power of CBIR systems.The results of this project will be available to the research community in this area, and will be integrated to the HRIS system under development between the Grupo de Base de Dados e Imagens (GBdI) of the ICMC-USP, and the Centro de Ciências de Imagens e Física Médica (CCIFM) of the Hospital das Clínicas de Ribeirão Preto (HCRP) - USP.