WEKO3
アイテム
Creating DICOM Structured Reporting Object from Free Text Reports Using the Text Mining Method
https://repo.qst.go.jp/records/68494
https://repo.qst.go.jp/records/6849440900699-7e97-4aff-b3fd-b2b975eac7af
Item type | 会議発表用資料 / Presentation(1) | |||||
---|---|---|---|---|---|---|
公開日 | 2006-06-19 | |||||
タイトル | ||||||
タイトル | Creating DICOM Structured Reporting Object from Free Text Reports Using the Text Mining Method | |||||
言語 | ||||||
言語 | eng | |||||
資源タイプ | ||||||
資源タイプ識別子 | http://purl.org/coar/resource_type/c_c94f | |||||
資源タイプ | conference object | |||||
アクセス権 | ||||||
アクセス権 | metadata only access | |||||
アクセス権URI | http://purl.org/coar/access_right/c_14cb | |||||
著者 |
Ando, Yutaka
× Ando, Yutaka× Tsukamoto, Nobuhiro× Kawaguchi, Osamu× Kasamatsu, Tomotaka× et.al× 安藤 裕× 塚本 信宏× 川口 修 |
|||||
抄録 | ||||||
内容記述タイプ | Abstract | |||||
内容記述 | Purpose:DICOM Structured Reporting has advantages, however, many existing reporting systems create and store in free-text format. In this study, we developed and evaluated the automated method for creating structured reports from free-text format. Methods:100 NM reports (internal data-set)of brain perfusion scintigraphy were analyzed and categorized into expression units using text-mining technology to create a dictionary. Using this dictionary, other 100 reports (external data-set)were analyzed for the accuracy. The results were also compared against manual categorization by physicians. Results:In internal data-set, 79.5% of sentences has matched, and in external data-set 62.2% of sentences has matched to convert into units. Discussion:Matching rate can be increased by improving the dictionary. In addition, by automatically creating DICOM structured reports, clinical information can be easily translated into other languages, transferred to other systems, and can be searched rapidly. |
|||||
会議概要(会議名, 開催地, 会期, 主催者等) | ||||||
内容記述タイプ | Other | |||||
内容記述 | RSNA'05 91th Scientific Assembly and Annual Meeting | |||||
発表年月日 | ||||||
日付 | 2005-12-02 | |||||
日付タイプ | Issued |