請用此 Handle URI 來引用此文件:
http://tdr.lib.ntu.edu.tw/jspui/handle/123456789/55326
標題: | 腫瘤超音波影像判別系統之改善與臨床應用 Enhancement for Computer-Aided Sonographic Tumor Classification Systems and its Clinical Application |
作者: | Hung-Lun Chang 張紘綸 |
指導教授: | 陳正剛 |
關鍵字: | 乳房腫瘤超音波影像,超音波影像診斷難易度評量,診斷表現評量, Breast Cancer Sonogram,Ultrasound Image Evaluation,Diagnostic Performance Assessment, |
出版年 : | 2014 |
學位: | 碩士 |
摘要: | 超音波檢查為早期乳房腫瘤最有效的非侵入性篩檢方法之ㄧ,專業醫師藉由觀察醫學影像之特徵提出持續追蹤該腫瘤或進一步接受細胞學檢查之建議。由於目前超音波影像皆為醫師憑藉著自身經驗及臨床觀察做出診斷,而資歷經驗不同的醫師對於相同的影像常有不同解讀而產生差異明顯的診斷結果,也會因為腫瘤影像的不同特性及難易度而有診斷結果差異。本研究利用專業醫療人員對於乳房腫瘤超音波影像的不同判讀結果,分析乳房腫瘤超音波影像及專業醫療人員判讀的統計特性,進一步提出乳房腫瘤超音波影像之臨床診斷難易度量化方法,並針對不同大小的腫瘤分析其診斷難易度。本研究同時分析專業醫療人員對於超音波影像判讀表現,藉由個別人員判讀結果與主流判讀結果的比較、與案例良惡性的比較、案例量化難易度、及BIRADS分級表現,提出評量醫師對於乳房超音波影像判讀表現的方法。
本研究利用臺大醫院100個臨床乳房腫瘤超音波影像案例及台灣乳房醫學會63位專業醫療人員及一套電腦輔助診斷系統軟體所提供之臨床BIRADS診斷結果,評估並驗證所提出之診斷難易度及診斷表現評量方法,利用本研究所提出的評量方法,未來可以該100臨床案例為標準提供做為乳房腫瘤超音波影像臨床案例判讀的訓練及診斷表現的評量。 The ultrasound imaging is one of the most effective non-invasive screening tools for early tumor diagnosis. Based on the characteristics observed on the medical images, clinicians make recommendations for patients to keep track of the tumors or undergo further cytology tests. However, due to clinicians’ experience difference and case variety, discrepancies in the diagnosis results of ultrasound imaging are very often observed. This research first uses the diagnosis differences made by clinicians with varied experiences to analyze the statistical properties of the cases and clinicians varieties. The quantitative methods to evaluate the degree of case difficulties are then proposed based on the discrepancy and diversity of the diagnostic results. Furthermore, this research attempts to quantitatively assess the clinician diagnosis performance based on their diagnosis made on the medical sonographies. Based on the diagnosis agreement with the majority diagnosis results and with the malignancy of the cases, assessment methods are proposed by considering both the degree of case difficulty and the clinician’s BIRADS classification performance. This research uses 100 cases of clinical ultrasound images collected in National Taiwan University Hospital and 63 medical professionals of Breast Cancer Society of Taiwan (BCST) to validate the proposed methods of case evaluation and clinician diagnosis assessment. With the proposed methods, the 100 cases can be used as a standard test set in the future for clinician training and performance assessment. |
URI: | http://tdr.lib.ntu.edu.tw/jspui/handle/123456789/55326 |
全文授權: | 有償授權 |
顯示於系所單位: | 工業工程學研究所 |
文件中的檔案:
檔案 | 大小 | 格式 | |
---|---|---|---|
ntu-103-1.pdf 目前未授權公開取用 | 45.43 MB | Adobe PDF |
系統中的文件,除了特別指名其著作權條款之外,均受到著作權保護,並且保留所有的權利。