REVIEW OF MEDICAL IMAGE RETRIEVAL SYSTEMS AND FUTURE DIRECTIONS

Abstract

The goal of this study is to present an overview of online systems for content-based medical image retrieval, with a focus on the United States (CBIR). The authors of this study hope to identify the advantages and disadvantages of these systems, as well as approaches to improving the relevance of multi-modal (text and picture) information retrieval in the iMedline system, which is currently under development at the National Library of Medicine, by the end of the study (NLM). A total of seven medical information retrieval systems were investigated in this study, including Figuresearch, BioText, GoldMiner, Yale Image Finder, Yottalook, Image Retrieval for Medical Applications (IRMA), and iMedline. Figuresearch was the most popular system among participants. The systems were assessed in accordance with the system of gaps described in [1. However, not all of these systems make advantage of the visual information supplied in biological literature in the form of figures and drawings, but a significant number of them do. All, on the other hand, make an attempt to extract image information from the full- text of the articles and to acquire figures and photos in response to a search query, which is a common practise. It is the purpose of iMedline to advance the state-of-the-art in multimodal information retrieval by merging image and text data in the calculation of relevance, a goal that has so far been accomplished. In this work, we discuss the shortcomings of current medical image retrieval systems, as well as future directions and next phases in the development of iMedline's context-based medical image retrieval system.