動態乳房彈性超音波之最佳影像選取與腫瘤診斷分析

Slice Selection and Diagnosis of Dynamic Breast Elastography

徐佑維(Yu-Wei Hsu)

指導教授 ： 張瑞峰

國立臺灣大學/電機資訊學院/資訊工程學研究所/碩士(2011年)

https://doi.org/10.6342/NTU.2011.03229

摘要

乳癌一直是全球婦女的主要死因之一，而且良性和惡性腫瘤間的不同硬度已經被醫生視為在觸診的重要特徵。近年來，乳房彈性超音波已經被使用來評估病人腫瘤的彈性程度，醫生需要對腫瘤組織施與輕微的壓力以便取得一段動態彈性視訊，而軟硬資訊則根據腫瘤組織的位移得到，另外，因為健康的組織和患病的組織有不同的軟硬程度，所以由乳房彈性超音波取得的彈性資訊已經被證明對於區分良惡性腫瘤是有幫助的，診斷時醫生將會從視訊中選出一張具有代表性的影像進行腫瘤分析。此篇論文的目的是利用提出的自動選圖方法選出一張影像並且針對這張影像來擷取特徵去診斷腫瘤。首先，為了減少不同醫生的主觀性選圖的影響，我們會利用我們提出的自動選圖方法來選出一張最具代表性的彈性影像。接著，會使用Level Set方法來自動地切割出腫瘤輪廓，而不是藉由醫生來手動圈選出腫瘤以保持切割結果的一致性。最後，我們會藉由腫瘤輪廓來擷取彈性特徵去診斷腫瘤。本實驗中以80個經過病理驗證的病例進行測試，包含45個良性以及35個惡性的病例，並且比較自動選圖方法所選影像和醫生選影像的診斷結果。經由實驗可以得知，我們提出的選圖方法的準確率為71.25%，靈敏度為91.43%，專一性則為55.56%;然而，當使用醫生選圖時，準確率只有為65.00%，靈敏度為77.14%，專一性則為55.56%。雖然自動選圖的靈敏度和準確率比醫生選圖好，但根據實驗結果統計分析，二者尚未具有統計上的差異。不過因自動選圖方法和醫生選圖在統計上是有相近的診斷結果，所以我們提出的自動選圖方法是可以幫助醫生選出具有代表性的影像以減少醫生選圖的時間。

關鍵字

彈性超音波 ； 乳房腫瘤 ； 腫瘤切割 ； 代表性影像 ； 電腦輔助診斷

並列摘要

The breast cancer is always the main causes of death for women and different firmness of benign and malignant tumors has been treated as an important characteristic by physicians during breast palpation. In recent years, the sonoelastography has been applied to evaluate the tumor strain of patient in clinical diagnosis. The physicians need to slightly press the tumor to obtain the dynamic elastographic image sequences. The tumor strain will be acquired on elastography based on the displacement of the tumor. Because the healthy and diseased tissues have different strain information, the elasticity information provided by the elastographic image in sonoelastography video has been proved to be useful in differentiating benign and malignant tumors. The physicians will select a representative slice from the dynamic elastographic image sequences to diagnose the tumor. In this study, the main purpose is to develop an automatic slice-selection method to select the representative slice from the sonoelastography video and then to diagnose the tumor by means of the elastographic features generated from the selected slice. Firstly, the representative slice is automatically selected by the proposed slice-selection method in order to reduce the selection variability of physicians. Then, the contour of tumor segmented by the physicians is substituted with an automatic segmentation of level set method so as to improve the consistency of the segmentation between the different operators. Finally, the contour of tumor is used to compute the elastographic features for diagnosing the breast tumor. This study has collected 80 biopsy-proved breast tumors comprised of 45 benign and 35 malignant lesions to estimate the performance of the slice-selection method. The representative slice chosen by the proposed scheme will be compared with the physician-selected slice. The experiment shows that the diagnosis performances of accuracy, sensitivity, and specificity evaluated by the leave-one-out method based on the elastographic features for the representative slice selected by the proposed slice-selection method are 71.25%, 91.43% and 55.56%, whereas 65.00%, 77.14% and 55.56% for the physician-selected slice. That is, the sensitivity and accuracy of proposed slice-selection method is better than physician-selected slice and the specificity of these two different schemes is similar. According to the statistical analysis of experimental result, the performance of the proposed slice-selection method is similar with that of the physician’s selection. Therefore, the proposed slice-selection method could assist the physician in selecting the appropriate representative slice and decreasing the time of selection.

並列關鍵字

sonoelastography ； breast tumor ； tumor segmentation ； representative slice

參考文獻

[24] R. C. Gonzalez, et al., Digital image processing, third ed. Upper Saddle River, New Jersey: Pearson Prentice Hall, 2009.

[1] D. M. Regner, et al., "Breast lesions: evaluation with US strain imaging--clinical experience of multiple observers," Radiology, vol. 238, pp. 425-37, Feb 2006.

[2] T. J. Hall, et al., "In vivo real-time freehand palpation imaging," Ultrasound in Medicine and Biology, vol. 29, pp. 427-35, Mar 2003.

[3] J. Ophir, et al., "Elastography: a quantitative method for imaging the elasticity of biological tissues," Ultrasonic Imaging, vol. 13, pp. 111-34, Apr 1991.

[4] B. S. Garra, et al., "Elastography of breast lesions: Initial clinical results," Radiology, vol. 202, pp. 79-86, Jan 1997.