目前在各醫療院所最常見的資料搜尋方式係採用關鍵詞搜尋法，此法最大的缺點之一在於資料儲存前未依病灶特徵做分類，以致產生搜尋結果資料過多或遺漏的問題。本研究為應用本體論的技術特點，以台灣南部某醫學中心的甲狀腺疾病影像判讀報告為例，分析其知識內容，將隱藏於報告中的專家知識，依疾病特徵及器官之間相互的關聯性加以分類，並標示腫瘤或器官等各類別的不同的屬性，建立判讀報告知識庫，透過它來描述複雜的知識及其問相互的關聯；並利用語意分析的方式將知識庫建置於Stanford大學Medical Informatics Center提出的Protégé知識平台上。本研究之知識內容層次的分析為著重在標準化及系統間的語意溝通，由使用者描述抽象問句，透過推論機制系統即可由知識架構中，有效找出符合語意的文件。

Most medical information systems in this country use keywords to search for relevant information. The major issue with this approach is the fact the keyword matching does not take into consideration the intended semantic of users, which may include elements of disease's symptoms, signs, and others. This study applies the ontology methodology to analyze image reports for thyroid disease according to its symptoms, tumor classifications, relations with other syndromes, and create a know ledge structure that interlink various disease elements in the database to facilitate relevant information retrieval. We, based on the actual cases from a major hospital, develop the know ledge taxonomy by analyzing the report contents starting with major disease symptoms and related body organs. The inter-relationships between symptoms and body organs were developed through attributes of various taxonomic classes. In building the system, we apply semantic network to analyze and develop relationships between knowledge items, and utilize the knowledge platform Protégé, developed by the Medical Informatics Center of Stanford University, for user to describe the abstract question. This can find out the file which accords with the user's purpose effectively from the know ledge structure through the inference mechanism.