請用此 Handle URI 來引用此文件:
http://tdr.lib.ntu.edu.tw/jspui/handle/123456789/69342
標題: | 應用於乳房超音波影像分類之電腦輔助診斷 Computer-aided Diagnosis for the Classification of Breast Ultrasound Images |
作者: | I-LING CHEN 陳怡伶 |
指導教授: | 張瑞峰 |
關鍵字: | 乳癌,超音波篩檢,腋窩淋巴結轉移,電腦輔助診斷,預測模型, Breast cancer,screening ultrasound,axillary lymph node metastasis,computer-aided diagnosis,prediction model, |
出版年 : | 2018 |
學位: | 博士 |
摘要: | 乳癌是目前女性癌症患者中第二大的死亡主因,隨著醫療持續發展,患者的五年存活率也大幅提升;然而,一旦發生癌細胞轉移,病灶擴散至腋窩淋巴結或是身體的其他器官,死亡率也將急遽上升。為此,許多篩檢工具以及診斷技術不斷革新,希望能在癌症初期盡早發現腫瘤,並能及時診斷,以進行最適當的治療策略。乳房超音波即是一項應運而生的重要篩檢技術,作為乳房攝影的輔助工具,超音波成像能發現相較於可觸診腫瘤更小體積、更為早期的可疑變異組織;許多的電腦輔助偵測和診斷系統,也應用大量蒐集的超音波影像搭配影像處理技術,發展更有效率的標準化腫瘤評估程序。儘管如此,許多作為乳癌診斷依據的影像特徵,隨著腫瘤體積的發展,才會愈趨具有區辨能力,因此,要在癌症初期做出正確的診斷並對症下藥,仍然是一件相當困難的事。為了解決分類學習過程中,小腫瘤的辨識特徵效能不佳的問題,本研究特別針對篩檢所取得的超音波影像資料集,發展電腦輔助診斷系統,並設計了依據腫瘤大小進行分類器分流建構的過濾器,有效提升診斷腫瘤良、惡性的自動分類準確度;進一步,更利用在乳房超音波所取得的腫瘤量化特徵,發展淋巴結轉移的自動預測模型。由於腋窩淋巴結狀態是用來評判是否發生乳癌轉移的至關要素,這項突破性的研究,特別依據乳房影像報告與資料系統所定義的腫瘤描述,將臨床上用來進行癌症分期的特徵,進行完整的特徵分析,並藉由腫瘤特徵取代腋窩淋巴結特徵,發展自動評估轉移風險的輔助工具。隨著乳癌早期檢測的推展,一些用以診斷轉移的腋窩淋巴結特性也隨之減少,放射科醫師在發現可疑組織的當下,將可利用本研究所開發的電腦輔助診斷系統,進行腫瘤良、惡性的分類,並在手術前取得轉移可能發生的風險評估,給予患者最佳的治療策略,達到早期診斷早期治療的目標。 Breast ultrasound (US) as a supplement to mammography has been used to verify the diagnostic ability for palpable lesions, and computer-aided diagnosis (CAD) techniques have been developed with various features extracted from US images. However, the diagnostic values of partial features are only feasible for predicting malignancy in tumors larger than 1 cm. An adaptive CAD model based on tumor size was proposed to classify breast tumors detected at screening US images using quantitative morphological and textural features. In addition, axillary lymph node (ALN) status is a key indicator in assessing and determining the treatment strategy for patients with newly diagnosed breast cancer. Previous studies suggest that sonographic features of a primary tumor have the potential to predict ALN status in the preoperative staging of breast cancer. In this study, a computer-aided prediction (CAP) model as well as the tumor features for ALN metastasis in breast cancers were developed using breast US images. In the present study, a screening database containing 156 tumors (78 benign and 78 malignant) was separated into two subsets of different tumor sizes (<1 cm and ≥1 cm) to explore the improvement in the performance of the CAD system for breast cancer. The other database containing 249 malignant tumors was acquired to test the differences between the non-metastatic (130) and metastatic (119) groups based on various features. Experiments show the CAD system can be helpful to classify breast tumors detected at screening US by the adaptive CAD model, as well as may useful for determining the ALN status in patients with breast cancer by the CAP model. |
URI: | http://tdr.lib.ntu.edu.tw/jspui/handle/123456789/69342 |
DOI: | 10.6342/NTU201801178 |
全文授權: | 有償授權 |
顯示於系所單位: | 資訊工程學系 |
文件中的檔案:
檔案 | 大小 | 格式 | |
---|---|---|---|
ntu-107-1.pdf 目前未授權公開取用 | 1.7 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。