图像分析在卵巢肿瘤中的鉴别意义

Differential Diagnosis of Borderline and Well-differentiated Mucinous Cystadenocarcinomas of the Ovary by Quantitative Nuclear Morphology

采用图像分析对 2 1例交界性粘液性囊腺瘤和 2 7例高分化粘液性囊腺癌进行形态定量分析和核仁组成区 ( Ag NOR)颗粒计数测定 ,建立分类函数。结果显示 :核体密度、核偏度、积分光密度、Ag NOR颗粒计数和核面积为两者鉴别的有效指标。在回顾性考核中 ,判别交界性粘液性囊腺瘤的敏感性为 95 % ,特异性为 10 0 % ;高分化粘液性癌的敏感性为 10 0 % ,特异性为 95 % ,其判别总错误率为 2 %。在前瞻性考核中 ,判别交界性粘液性囊腺瘤的敏感性为 86% ,特异性为 91% ;高分化粘液性癌的敏感性为 91% ,特异性为 86% ,其判别总错误率为 11%。提示 :定量病理学方法在两者鉴别诊断中具有有效性和实用价值。

Multiple descriptors extracted from nuclear tracing and intranuclear gray level analysis of argyrophilic nucleolar organizer regions were quantitatively evaluated in 21 borderline and 27 well differentiated mucinous cystadenocarcinomas by image analysis technology.The results showed the parameters of nuclear area,integrated optical density and AgNOR number and histological structure (LPP and Vv) were significant different between the groups,and used to develop a classification function.In the development set,these 2 discriminant functions demonstrated respective sensitivity and specificity of 95% and 100% for borderline,100% and 95% for well differentiated mucinous cystadenocarcinomas.In the test set,the sensitivity and specificity were 86% and 91% for borderline,91% and 86% for well differentiated mucinous cystadenocarcinomas.Sum of error rate was only 2% in retrospective validation,but 11% in prospective validation.Quantitative nuclear morphology may be useful in discrimination between these tumors.