简化MRI联合超声S-Detect模型对乳腺肿块良恶性鉴别的价值

Value of the model constructed by abbreviated MRI combined with ultrasound S-Detect for the differentiation of benign and malignant breast masses

目的探究简化MRI联合超声S-Detect模型对乳腺肿块良恶性鉴别的价值。方法选取华北理工大学附属唐山市妇幼保健院2021年3月~2023年1月行乳腺超声和MRI检查的154例患者(159个病灶)作为研究对象。以病理活检结果为金标准。简化MRIⅠ为乳腺影像报告和数据系统(BI-RADS)分类模型;简化MRIⅡ将BI-RADS分类4A及以下定义为良性,将BI-RADS分类4B及以上定义为恶性。采用Kappa检验分析不同方法鉴别乳腺肿块良恶性结果与病理结果的一致性;采用Logistic回归构建简化MRIⅠ和超声S-Detect+简化MRIⅠ鉴别乳腺肿块良恶性的模型;采用ROC曲线和决策曲线分析评价不同方法鉴别乳腺肿块良恶性的价值。结果病理结果显示,乳腺肿块中良性43例,恶性116例。超声S-Detect+简化MRIⅠ鉴别乳腺肿块良恶性的准确率高于超声S-Detect(P<0.05),与简化MRIⅠ、简化MRIⅡ和超声S-Detect+简化MRIⅡ的准确率差异无统计学意义(P>0.05)。超声S-Detect+简化MRIⅠ的Kappa值高于超声S-Detect、简化MRIⅠ、简化MRIⅡ和超声S-Detect+简化MRIⅡ。超声SDetect+简化MRIⅠ鉴别乳腺肿块良恶性的ROC曲线下面积高于超声S-Detect、简化MRIⅠ、简化MRIⅡ和超声S-Detect+简化MRIⅡ(P<0.05)。决策曲线分析结果显示,在全风险阈值范围内,超声S-Detect+简化MRIⅠ鉴别乳腺肿块良恶性的净收益高于超声S-Detect和简化MRIⅡ;在绝大部分风险阈值范围内,超声S-Detect+简化MRIⅠ鉴别乳腺肿块良恶性的净收益高于简化MRIⅠ和超声S-Detect+简化MRIⅡ。结论简化MRI联合超声S-Detect模型有助于乳腺肿块良恶性鉴别,其价值高于单纯简化MRI和超声S-Detect。

Objective To investigate the value of abbreviated MRI combined with ultrasound S-Detect model for benign and malignant differentiation of breast masses.Methods A total of 154 patients(159 lesions)who underwent breast ultrasound and MRI from March 2021 to January 2023 at Tangshan Maternal and Child Health Hospital affiliated to North China University of Science and Technology were selected as study subjects.Pathologic biopsy results were used as the gold standard.The abbreviated MRIⅠrepresented the Breast Imaging Reporting and Data System(BI-RADS)classification model.For abbreviated MRIⅡ,BI-RADS classifications of 4A and below were deemed benign,while classifications of 4B and above were deemed malignant.The consistency between the results of different methods for identifying benign and malignant breast masses and pathology results was analyzed with the Kappa test.Logistic regression was used to construct models for identifying benign and malignant breast masses by abbreviated MRIⅠand ultrasound S-Detect+abbreviated MRIⅠ.The value of different methods to identify benign and malignant breast masses was evaluated using the ROC curve and decision curve analysis.Results Pathologic findings showed 43 benign and 116 malignant breast masses.The accuracy of ultrasound SDetect+abbreviated MRIⅠin identifying benign and malignant breast masses was higher than that of ultrasound S-Detect alone(P<0.05).It was also comparable to the accuracy of abbreviated MRIⅠ,abbreviated MRIⅡ,and ultrasound S-Detect+abbreviated MRIⅡ(P>0.05).The Kappa value of ultrasound S-Detect+abbreviated MRIⅠwas higher than that of ultrasound S-Detect,abbreviated MRIⅠ,abbreviated MRIⅡ,and ultrasound S-Detect+abbreviated MRIⅡ.The area under the ROC curve for identifying benign and malignant breast masses with ultrasound S-Detect+abbreviated MRIⅠwas higher than that with ultrasound S-Detect,abbreviated MRIⅠ,abbreviated MRIⅡand ultrasound S-Detect+abbreviated MRIⅡ(P<0.05).Decision curve analysis results showed that within the full risk threshold,the net benefit of identifying breast masses with ultrasound S-Detect+abbreviated MRIⅠwas higher than that with ultrasound S-Detect and abbreviated MRIⅡ.In the vast majority of the risk threshold range,the net benefit of ultrasound S-Detect+abbreviated MRIⅠin identifying benign and malignant breast masses was higher than that of abbreviated MRIⅠand ultrasound S-Detect+abbreviated MRIⅡ.Conclusion The model constructed by abbreviated MRI combined with ultrasound S-Detect can help identify benign and malignant breast masses with a higher value than abbreviated MRI and ultrasound S-Detect alone.