案例推理法增進乳癌診斷率

Case-based Reasoning to Improve Diagnostic Accuracy in Breast Cancer

張偉斌(Wei-Pin Chang) ； 吳振龍(Chen-Long Wu) ； 紀櫻珍(Ying-Chen Chi) ； 黃育文(Yu-Wen Huang) ； 劉德明(Der-Ming Liou)

《北市醫學雜誌》 3卷11期 (2006/11) Pp. 78-84

https://doi.org/10.6200/TCMJ.2006.3.11.08

摘要

目的：根據94年衛生署公佈在國人十大死因中，女性乳癌高踞台灣所有癌症死亡原因第四名，且是死亡年齡最年輕的癌症，這些被乳癌奪去生命的女性，平均年齡約為55歲。因乳癌死亡人數佔所有癌症死亡之人數也不斷的上升，從民國83年死亡人數788人至民國93年升高為1339人，增加551人，成為台灣地區女性之一大威脅。方法：本研究以美國Wisconsin大學附設醫院所提供的乳癌病患資料為基礎，資料內容的九個變項(variables)都是根據病理報告(pathology report)所摘錄出，共有699筆乳癌患者的資料，每筆均有描述病人之癌症結果為何。結果：此資料內容根據病理症候之嚴重程度給予1-9分的數值(value)，因此在本研究中將透過案例推理機制(Case-based Reasoning)的方式，以歐基里德距離(Euclidean Distance)計算方式為基礎，計算由使用者輸入案例與存放案例庫中既有案例的相似性(similarity)，由既有之案例推衍現今病患之結果，將有利醫療人員及早做出正確的醫療診斷。結論：醫療人員可以透過瀏覽器操作此案例推理系統，也可透過本系統新增案例、刪除案例等，因而本研究系統設計為具有回饋性之系統，可透過不斷增加相同之病患資料，而提升系統之分類準確度，期望能藉由此系統成為醫療人員進行診斷或是從事醫學研究的助力。

關鍵字

資料探勘 ； 案例式推理 ； 乳癌

並列摘要

Background and Purpose: In Taiwan, breast cancer is an issue of particular importance in public health, and it has been ranked one of the most common cancers affecting Taiwanese women for many years. According to Taiwan's Department of Health statistics, breast cancer is one of the cancers that has seen the steepest increase in the number of cases between 1994 to 2004. Approximately half of all breast cancers affect women under the age of 55, one quarter occur in women between the ages of 55 and 69, and the remaining cases develop in women who are 70 years old or above. In 2004, 1,339 Taiwanese women died of breast cancer, the fourth leading cause of cancer death in Taiwanese women. Methods: To reduce the high number of biopsies performed on benign lesions, a case-based reasoning (CBR) classifier was developed to predict biopsy results. In this study, the standard reference was based on 699 breast cancer cases at the University of Wisconsin Hospital, USA. Results: According to different pathological features, each case was classified from 1~9 in our database. We defined CBR by using Euclidean distances to determine the similarity between the test case and the reference cases. The CBR system may be updated by medical staff through browsers to add or delete patient information on the reference database, and thereby improve accuracy of classification. Conclusion: Such a system may provide advantages in terms of improved diagnostic accuracy and medical research.

並列關鍵字

data mining ； case-based reasoning ； breast cancer