Social Networking Service for Development of Multilingual RadLex of East Asian Languages Linked with Radiology Image Search Engines
 
Authors:
Norio Nakata, MD, Jikei University, School of Medicine; Kunihiko Fukuda, MD
 
Background:
Radlex is an effort of RSNA to develop a globally accepted lexicon for radiology of standardized terms. Recently some projects of multilingual Radlex project of German, Spanish, and Portugal) have already started in the world. We have developed a work-in-progress anatomic term browser based on multilingual (Japanese, Chinese, Korean, and English) Radlex. RadLex will be basic resources for radiological structured reporting.

International teleradiology has been started and its global expansion of scale is also expected. Multilingual RadLex is thought to be very useful for global scale teleradiology in the near future. However, there are some local language issues, especially in East Asian countries such as China and Japan. One of the solutions is structured reporting using semi-automatic translation system. In this paper, we tried to translate anatomic and imaging observation characteristic terms from original English to Japanese, Chinese, and Korean. It consisted of 421 terms from a main anatomic element. The RadLex API acts as a web service, which allowed developers to query the RadLex terminology and retrieve an XML response.
 
Evaluation:
Some multilingual technical terms have differences of culture, expression method, and rhetoric. Chinese and Japanese languages, which consist of Kanji characters, are totally different from alphabetical languages, such as English. So we have constructed the open source Social Networking Service (SNS) website for Multilanguage translation of RadLex in order to solve various problems in translation. The open source SNS system was developed by BuddyPress (http://buddypress.org/), which is a plug-in, took of WordPress (http://wordpress.org/). We use Protégé for making XML database of multilingual RadLex. This year we focused on “imaging observation characteristics” of RadLex. In this part, there are some specific English expressions, such as “patchy” or “impinged,” which are difficult to translate directly from English to Japanese Kanji. In our SNS, some Japanese and Chinese users have discussed about those translation problems. For example, we have discussed using borrowed foreign words for those difficult English terms. For educational purposes, we have also created the pages of RadLex terms of “imaging observation characteristics” linked with radiology image search engines, such as ARRS GoldMiner and Yottalook. We also changed the web design for mobile devices, such as iPhone and smart phone, as a convenience of radiological residents and medical students.
 
Discussion:
The contents of radiological diagnostic reporting should be understood not only by radiologists, physicians, but also by patients. Usually patients read their own information, including diagnostic reports, using native language. So multilingual translation is very important for this system. We have continued to make multilingual Radlex of Asian languages. We encountered some problems of translation. SNS is one of the solutions of such problems.

Through this project, translation of RadLex “imaging observation characteristic” terms has advantage of radiology education for younger Japanese radiologists. They have learned how to use such technical terms in their reading reports. Direct links of ARRS GoldMiner are also helpful for this project.

A multilingual structured reporting system will be progressed the global teleradiology framework as a point of view of software. One important concept is an open attitude for a variety of worldwide radiologists, easy accessible environment of discussion web space. There are fewer contents of Chinese or Japanese languages about advertisement of RadLex. Not many Asian radiologists know about the RSNA RadLex project, with a few exceptions that are interested in informatics in radiology. So we have to continue to spread the information about RadLex the same as the United States. However, there are some grammatical and linguistic problems in many world languages like Chinese and Japanese. European languages, such as German and Spanish, are identical origin of Indo-European languages. But Chinese and Japanese languages are totally different from Indo-European languages in characters, phonology, morphology, syntax, lexis, semantics, and pragmatics. So we have to continue to discuss many technical terms in case of translation efforts. Because Korean Hangeul words have many homophones, RadLex translation is sometimes difficult.

Another aspect of multilingual RadLex is professional education of Chinese and Japanese young radiologists and medical students. 20-30 years ago many Japanese radiologists wrote the reading reports using an English typewriter. But now Japanese patients recognize the importance of their own medical information, including reading reports written by radiologists. So recently almost all Japanese radiologists write their reading reports using the Japanese language. Multilingual RadLex is an important basic educational resource to learn radiological technical terms in English and to match Japanese words. Medical students and residents are sometimes unfamiliar with radiological technical terms, especially imaging observation characteristic terms. So our website can demonstrate typical diagnostic images of imaging observation characteristic terms using radiology image search engines automatically.
 
Conclusion:
This project will be useful for international teleradiology using multilingual structured reporting in the near future.