How It Works

Unlike most internet search engines, ARRS GoldMiner® understands medical vocabulary. It uses sophisticated techniques from the U.S. National Library of Medicine (part of NIH) to discover medical concepts in free-text figure captions, and uses that information to quickly retrieve relevant images. GoldMiner® incorporates standardized vocabularies, such as the Medical Subject Heading (MeSH) terms, which are used to index the medical literature in MEDLINE and PubMed.

ARRS GoldMiner® recognizes abbreviations, synonyms, and kinds of diseases. Not only does it know that "renal calculi" and "kidney stones" mean the same thing, it knows that renal calculi are a type of kidney disease.

GoldMiner searches by both concepts and keywords. As a result, searches for "renal calculi" and "kidney stones" won't find exactly the same entries. If a figure caption says "stones are seen in the kidney", the words "stones" and "kidney" will be indexed, but not necessarily the concept of kidney stones / renal calculi.

As with a conventional search engine, GoldMiner's search results can depend on the presence of specific words in the figure captions.

Using ARRS GoldMiner®

To find images, enter a search term that describes a disease, imaging finding, or anatomic feature.

You can filter the search results by imaging modality (e.g., "CT"), age, and/or sex - just click on the pull-down menus in the grey bar at the top of the search results page. The selections will indicate the number of images for each selection.

Click on the small "thumbnail" image to open a full-size image on the original journal's web site. You can link to the full text of the article as well - just click on the article's name.

Vocabularies

ARRS GoldMiner uses the Unified Medical Language System (UMLS®) Metathesaurus® from the U.S. National Library of Medicine.

This material includes SNOMED Clinical Terms® (SNOMED CT®) which is used by permission of the International Health Terminology Standards Development Organisation (IHTSDO). All rights reserved. SNOMED CT®, was originally created by The College of American Pathologists. "SNOMED" and "SNOMED CT" are registered trademarks of the IHTSDO.

The ARRS GoldMiner® Team

• Michael Bruno
• Charles Kahn, chair
• Susanna Lee
• Henning Müller
• Michael Richardson
• Daniel Rubin
• Beverly Wood
• Hong Yu

• John Finney
• Gerald Cameron

Publications

1. A digital library of radiology images. AMIA Annual Symposium Proceedings 2006; 972. [Full Text] [PDF]
2. GoldMiner: a radiology image search engine. AJR 2007; 188:1475-1478. [Full Text] [PDF]
3. Effective metadata discovery for dynamic filtering of queries to a radiology image search engine. Journal of Digital Imaging 2008; 21:269-273. [Full Text] [PDF]
4. Dynamic "inline" images: Context-sensitive retrieval and integration of images into web documents. Journal of Digital Imaging 2008; 21:274-279. [Full Text] [PDF]
5. A multilingual image search engine. AMIA Annual Symposium Proceedings 2008; 995. [PubMed: 18998789]
6. Multilingual retrieval of radiology images. RadioGraphics 2009; 29:23-29. [Full Text] [PDF]
7. Automated semantic indexing of figure captions to improve radiology image retrieval. Journal of the American Medical Informatics Association 2009; 16:380-386. [Full Text] [PDF]
8. Ontology-assisted analysis of web queries to determine the knowledge radiologists seek. Journal of Digital Imaging 2011; 24:160-164. [Full Text] [PDF]