探讨S-Detectd的诊断效能及其临床应用价值。

选取2019年4月至7月于北京大学肿瘤医院乳腺中心超声室接受超声诊断及S-Detect检查的患者378例，共计390个病灶进行回顾性分析。以组织病理诊断为金标准，应用诊断试验四格表分别计算超声医师及S-Detect对乳腺病灶良恶性诊断的敏感度、特异度、准确性、阳性预测值及阴性预测值，采用Kappa检验对S-Detect及超声医师与病理诊断结果的一致性进行分析；对S-Detect判断不确定的病灶进行假设判定，假设判定均为恶性为S-Detect 1、均为良性为S-Detect 2，并分别计算S-Detect 1与S-Detect2的诊断敏感度、特异度、准确性、阳性预测值及阴性预测值。

病理结果提示，378例患者共计390个病灶中恶性病灶260个，良性病灶130个。S-Detect的诊断敏感度、特异度、阳性预测值、阴性预测值与准确性分别为94.6%、56.2%、81.2%、83.9%、81.8%。超声医师的诊断敏感度、特异度、阳性预测值、阴性预测值与准确性分别为100.0%、9.2%、71.0%、100.0%、69.7%。Kappa分析结果显示：S-Detect与病理诊断结果的一致性较超声医师与病理诊断结果的一致性好（Kappa值：0.553 vs 0.119，P<0.05）。将S-Detect判断不确定的69个病灶进行假设判定，结果显示，S-Detect 1的诊断敏感度、特异度、阳性预测值、阴性预测值和准确性分别为94.6%、56.2%、81.2%、83.9%、81.8%；S-Detect 2的诊断敏感度、特异度、阳性预测值、阴性预测值和准确性分别为79.6%、79.2%、88.5%、66.0%、79.5%。

S-Detect技术对乳腺病灶具有一定的诊断价值，尤其对于良性病灶的诊断符合率优于超声医师，但在临床应用中仍存在部分病灶判断不明确等局限性，如何在临床中更加恰当地结合使用S-Detect还需要更深入的数据研究加以证实。

To assess the diagnostic efficiency and clinical application value of S-Detect in breast lesions.

A total of 378 patients with 390 breast lesions who underwent ultrasound and S-Detect examinations at the Ultrasound Department of the Center for Breast Diseases of Peking University Cancer Hospital (Beijing Cancer Hospital) from April to July 2019 were selected for this retrospective analysis. Taking histopathological diagnosis as the gold standard, the four-fold table method was used to calculate the sensitivity, specificity, accuracy, positive predictive value, and negative predictive value of ultrasound and S-Detect for the diagnosis of benign and malignant breast lesions. The Kappa test was used to examine the consistency of the results of ultrasound and S-Detect with the pathological diagnosis. Hypothetical judgments were made on the lesions with an uncertain diagnosis by S-Detect, in which those judged as malignant and benign lesions were designated as S-Detect 1 and S-Detect 2, respectively, and the diagnostic sensitivity, specificity, accuracy, positive predictive value, and negative predictive value for S-Detect 1 and S-Detect 2 were calculated.

The pathological results showed that among the 390 lesions in the 378 patients, 260 were malignant lesions and 130 were benign lesions. The diagnostic sensitivity, specificity, positive predictive value, negative predictive value, and accuracy of S-Detect were 94.6%, 56.2%, 81.2%, 83.9%, and 81.8%, respectively; the corresponding values of ultrasound were 100.0%, 9.2%, 71.0%, 100.0%, and 69.7%. Kappa analysis showed that the consistency between S-Detect and pathological diagnosis results was better than that between ultrasound and pathological diagnosis results (Kappa value: 0.553 vs 0.119, P<0.05). Sixty-nine lesions with an uncertain diagnosis by S-Detect underwent hypothetical judgments. The results showed that the diagnostic sensitivity, specificity, positive predictive value, negative predictive value, and accuracy for S-Detect 1 were 94.6%, 56.2%, 81.2%, 83.9%, and 81.8%, respectively; the corresponding values for S-Detect 2 were 79.6%, 79.2%, 88.5%, 66.0%, and 79.5%.

S-Detect technology has appreciated diagnostic value for breast lesions; especially for benign lesions, the diagnostic performance is better than that of ultrasound. However, there are some limitations in clinical application such as unclear judgment of some lesions. How to use S-Detect more appropriately in clinical setting requires more in-depth research to confirm.