Jianguo Zhang PhD, Presenter: Nothing to Disclose

Tonghui Ling MS, Abstract Co-Author: Nothing to Disclose

Jianyong Sun, Abstract Co-Author: Nothing to Disclose

Suo Li, Abstract Co-Author: Nothing to Disclose

Yuanyuan Yang MS, Abstract Co-Author: Nothing to Disclose

Kai Zhang BS, Abstract Co-Author: Nothing to Disclose

We had designed a searching engine combining semantic space searching and CBIR techniques to search lung CT images with solitary pulmonary nodules (SPN) in PACS environment, and presented this prototype system in scientific presentation in 2009 RSNA Conference. In this presentation, we present new approach to extend this searching engine to cover more organs and lesions, gave an implementation of this searching engine in RIS-integrated PACS, and discussed its clinical evaluation.

The studies of cardiology CT images with coronary heart disease, brain CT and MR images with stroke, and abdomen CT with colorectal cancer as well as lung CT images with SPN, were included in our research. The diagnostic reports of the studies with findings of lesions are first analyzed by a NLP engine and then indexed in an inverted index. The contents of images of related to the reports are identified by their low level features extracted from the ROIs of images containing the lesions and indexed in a specified high-dimensional database. The first step in using this search engine uses the inverted index to search for relevant radiology report matching the symptoms or diagnoses specified by users as query criteria. The second step searches and retrieves the features of images from the high dimensional database associated with each report returned in the first step and computes the feature similarities between user query image and the retrieved images. The final search results are then sorted by similarities computed on the second step.

The developed searching engine was integrated a clinical RIS-integrated PACS, and operated for two years in Huadong hospital in Shanghai. There were about more than 30 cases averagely being searched and retrieved by using this searching engine daily. But the usages of this search engine for the purposes of decision support, research, and education were quite different.

The developed content-based searching engine can be easily integrated with a clinical RIS-integrated PACS and has been operating for two years in a hospital. The evaluation results showed that searching engine can be used for the purposes of decision support, research, and education.

The presented search engine is extremely useful to assist radiologists, medical researchers and students to mine meaningful information from PACS and RIS for their decision support, research and case-

Zhang, J, Ling, T, Sun, J, Li, S, Yang, Y, Zhang, K, Implementation and Clinical Evaluation of Content-based Searching Engine in RIS-integrated PACS.  Radiological Society of North America 2013 Scientific Assembly and Annual Meeting, December 1 - December 6, 2013 ,Chicago IL. http://archive.rsna.org/2013/13018226.html