Yutaka Ando MD, Presenter: Nothing to Disclose

Osamu Kawaguchi, Abstract Co-Author: Nothing to Disclose

Hikaru Futami, Abstract Co-Author: Nothing to Disclose

Hiromasa Yamagishi, Abstract Co-Author: Nothing to Disclose

Hirofumi Fujii MD, Abstract Co-Author: Nothing to Disclose

Nobuhiro Tsukamoto MD, Abstract Co-Author: Nothing to Disclose

Tomotaka Kasamatsu MD, Abstract Co-Author: Nothing to Disclose

Hiroshi Kaneko, Abstract Co-Author: Nothing to Disclose

Masakazu Osada PhD, Abstract Co-Author: Employee, Toshiba Corporation

Kaoru Kurosaki, Abstract Co-Author: Employee, Toshiba Corporation

et al, Abstract Co-Author: Nothing to Disclose

et al, Abstract Co-Author: Nothing to Disclose

Usually diagnostic reports are described in free-text format. But, it is difficult to search quickly and accurately clinical cases using the format. Consequently it is necessary to structuralize them semantically. In this study, we proposed and evaluated the structuring method of diagnostic reports for semantic interpretation.

We defined a description unit to convert free-text reports into the semantic information. Each description units consisted of 5 components; finding/diagnosis, modifier, region, regional modifier, and confidence. We developed the system to automatically extract description units from free-text reports referring to the dictionary. We verified this system by applying it to diagnostic reports of chest CT. We made dictionary by picking up the words of description units from reports that contain 500 cases. The validation reports were 100 cases (internal cases) selected from the 500 cases and another 100 cases (external cases). The interpretation was confirmed manually by the physician.

Our method was applicable to diagnostic reports of chest CT. The number of registration words of dictionary were 2587,and the matching rate of internal cases was 90.0%, that of external cases was 78.3％. The matching rate was tended to improve as the number of registered words of the dictionary increased.

We proposed the method to structuralize diagnostic reports for semantic interpretation by using description units, and we developed the system to extract them automatically. We confirmed that free-text reports were converted into structured reports with tuning the registration words for the dictionary.

The method that we proposed is expected that the free-text diagnostic reports could apply to storage of DICOM-SR format or searching similar clinical cases with semantic interpretation.

Ando, Y, Kawaguchi, O, Futami, H, Yamagishi, H, Fujii, H, Tsukamoto, N, Kasamatsu, T, Kaneko, H, Osada, M, Kurosaki, K, et al, , et al, , The Structuring Method of Diagnostic Reports by Using Description Units for Semantic Interpretation.  Radiological Society of North America 2007 Scientific Assembly and Annual Meeting, November 25 - November 30, 2007 ,Chicago IL. http://archive.rsna.org/2007/5014956.html