You are here

Open-i

Screen detail of OpenI Web page
Project information

The Open-ism (pronounced “open eye”) experimental multimedia search engine retrieves and displays structured MEDLINE citations augmented by image-related text and concepts and linked to images based on image features.

The Open-i project provides novel information retrieval services for biomedical articles from the full text collections such as PubMed Central. It is unique in its ability to index both the text and images in the articles. The article retrieval is powered by the LHNCBC-built search engine Essie.

The Open-I biomedical image search engine lets users retrieve not only the MEDLINE citation information, but also the outcome statements in the article and the most relevant figure from it. Further, it is possible to use the figure as a query component to find other relevant images or other visually similar images. Future stages aim to provide image region-of-interest (ROI) based querying. The initial number of images is projected to be around 600,000 and will scale to millions. The extensive image analysis and indexing and deep text analysis and indexing require distributed computing. Future plans include making the image computation services available as a NLM service.

Publications/Tools: 
Xue Z, Antani SK, Long LR, Demner-Fushman D, Thoma GR. Improving face image extraction by using deep learning technique. Proc. SPIE 9789, Medical Imaging 2016: PACS and Imaging Informatics: Next Generation and Innovations, 97890J (March 25, 2016); doi:10.1117/12.2216278.
Demner-Fushman D, Shooshan SE, Rodriguez L, Antani SK, Thoma GR. Annotation of Chest Radiology Reports for Indexing and Retrieval. Multimodal Retrieval in the Medical Domain 2015 (MRMD 2015), Vienna, Austria, March 29, 2015
Vajda S. Label the many with a few: Semi-automatic medical image modality discovery in a large image collection. IEEE Symposium Series on Computational Intelligence, Orlando, FL, Dec 9-12, 2014
Rahman MM, You D, Simpson MS, Antani SK, Demner-Fushman D. Interactive Cross and Multimodal Biomedical Image Retrieval Based on Automatic Region-Of-Interest (ROI) Identification and Classification. International Journal of Multimedia Information Retrieval (MMIR) 2014 Sep;3(3):131-146.
Kilicoglu H, Demner-Fushman D. Coreference Resolution for Structured Drug Product Labels Proceedings of BioNLP 2014. pp. 44-53.
You D, Antani SK, Demner-Fushman D, Thoma GR. Does Figure-Text Improve Biomedical Article Retrieval? A Pilot Study. CBMS 2014. May 2014.
Xue Z, Antani SK, Long LR, Demner-Fushman D, Thoma GR. Body Segment Classification for Visible Human Cross Section Slices. IEEE 27th International Symposium on Computer-Based Medical Systems (CBMS), New York, NY, May 2014: pp. 199-204
Xue Z, You D, Antani SK, Long LR, Demner-Fushman D, Thoma GR. Classification of Visual Signs in Abdominal CT Image Figures in Biomedical Literature. Proc. SPIE. 9039. Medical Imaging 2014: PACS and Imaging Informatics: Next Generation and Innovations. 90390T.
Chachra S, Xue Z, Antani SK, Demner-Fushman D, Thoma GR. Extraction and Labeling High-resolution Images from PDF Documents. Proc. SPIE 9021, Document Recognition and Retrieval XXI, 90210Q (December 27, 2013). doi:10.1117/12.2042336.
Xue Z, Antani SK, Long LR, Demner-Fushman D, Thoma GR. Classification of CT Figures in Biomedical Articles Based on Body Segments. 2013 IEEE International Conference on Healthcare Informatics (ICHI), pp.264,268, 9-11 Sept. 2013. doi: 10.1109/ICHI.2013.17.

Pages