結合超音波紋理特徵和血管分佈評估複雜性卵巢腫瘤的良惡性

目的： 卵巢癌在我國女性中的發生率和死亡率均排名第七，提高早期診斷能力對降低死亡率至關重要。然而，不同類型的卵巢腫瘤需有精確的診斷以避免過度治療，特別是對年輕患者的生育考量。歐美國家的研究常基於超音波特徵來分類腫瘤，但這些規則不完全適用於臺灣。為了改善診斷精確度，本研究結合紋理分析和彩色都卜勒影像，專注於分辨複雜性卵巢腫瘤，並開發電腦輔助診斷系統以協助醫師更精確地診斷。材料與方法： 本研究收錄88位受試者，根據對應病理切片報告分成良性組與惡性組，結合能夠代表組織微觀結構物理特性的超音波定量參數和能夠判斷血流分佈的彩色都卜勒結果，即紋理特徵參數和血管分佈評分，藉由結合這兩種參數對受試者的卵巢腫瘤進行分析，觀察良、惡性腫瘤對於組織微觀結構和血管分佈的差異是否會反映在超音波定量參數及彩色都卜勒結果上。結果：研究結果顯示，透過對卵巢腫瘤的紋理特徵參數進行接收者操作特徵曲線（ROC曲線）分析，發現在串聯這五個紋理參數的分析時，其診斷效能之曲線下面積、準確度、靈敏度與特異度分別為0.724、72.7%、51.4% 和88.2%，顯示出其高特異性。另外，單純只看血管分佈評分時，其診斷效能分別為0.786、79.5%、73% 和84.3%，顯示出其高敏感性。當我們將兩者結合時，發現整體診斷效能皆有所提升，分別為0.902、 84.1%、83.8%和84.3%，能同時展現出高靈敏度和高特異性，提供更準確的診斷。結論：本研究合併不同物理意義的參數，透過使用紋理特徵參數與血管分佈診斷，發現可以達到相輔相成的作用，彌補彼此之間的不足，並提升偵測良性及惡性卵巢腫瘤的能力。這項發現不僅為醫師提供了強而有力的診斷工具，還為未來的研究開闢了一條新途徑，可以結合不同的診斷維度，提高診斷的精確度。

Objective：Ovarian cancer ranks the seventh in both incidence and mortality among women in our country. Enhancing early diagnostic capabilities is essential for reducing mortality rates. However, precise diagnosis is necessary for different types of ovarian tumors to prevent overtreatment, particularly considering fertility issues for younger patients. While studies in Western countries often classify tumors based on ultrasound features, these guidelines are not fully applicable to Taiwan. To improve diagnostic accuracy, this study integrates texture analysis and color Doppler imaging, focusing on distinguishing complex ovarian tumors and developing a computer-assisted diagnosis system to aid physicians in more accurate diagnoses.Materials and methods：This study included 88 participants that were divided into benign and malignant groups and was based on corresponding histopathological reports. It integrated quantitative ultrasound parameters which represented the microscopic physical characteristics of tissues with the results from Color Doppler imaging which assessed blood flow distribution and was namely texture feature parameters and Location of Blood Vessels score. By combining these two parameters, the study analyzed ovarian tumors of participants and observed whether differences in tissue microstructure and vascular distribution between benign and malignant tumors were reflected in the quantitative ultrasound parameters and Color Doppler results or not.Results：The study results showed that through the Receiver Operating Characteristic （ROC）curve analysis of texture feature parameters of ovarian tumors, when analyzing the combination of these five texture parameters, the area under the curve, accuracy, sensitivity, and specificity were respectively 0.724, 72.7%, 51.4%, and 88.2% which indicated its high specificity. Additionally, when evaluating was solely based on Location of Blood Vessels score, the diagnostic performance respectively showed that the area under the curve, accuracy, sensitivity, and specificity were 0.786,79.5%, 73%, and 84.3% which demonstrated high sensitivity. When combining both methods, the overall diagnostic effectiveness improved, with values of 0.902, 84.1%, 83.8%, and 84.3%, thereby exhibiting both high sensitivity and high specificity, and providing more accurate diagnoses.Conclusion：This study combined the parameters of different physical significance through the use of texture feature parameters and Location of Blood Vessels score diagnostics, and found that they can complement each other and compensate for their respective deficiencies, enhancing the capability to detect benign and malignant ovarian tumors. This discovery not only provides physicians with a powerful diagnostic tool but also paves a new path for future research to integrate various diagnostic dimensions, thereby increasing diagnostic accuracy.