Image Retrieval for Open Access PubMedCentral Archive

摘要

Traditional literature search engine in biomedicine domain focus mainly on summary content such as title, abstract, keywords, journal published, author etc. There is a need to develop full-text search engine to encompass not only summary information, but also experimental details described in results and discussion. Images in the paper convey important and rich information about the experimental design and findings and are key components for results and discussion. Yet so far there is no image retrieval system implemented for literature databases. In this study, we build a prototype for image retrieval of PubMedCentral, an open access biomedical literature database containing full-text articles, to provide researchers a useful tool in looking into details of supporting evidences for scientific findings.