人工智能S-Detect技术联合弹性成像对乳腺良恶性结节的鉴别诊断价值

刘伯元; 李胜男

人工智能S-Detect技术联合弹性成像对乳腺良恶性结节的鉴别诊断价值

Significance of artificial intelligence S-Detect technology combined with elastography for the differential diagnosis of benign and malignant breast nodules

摘要

摘要:
目的： 探讨人工智能S-Detect技术联合超声弹性成像（UE）对乳腺良恶性结节的鉴别诊断价值。
方法： 采用S-detect、UE及常规超声（CUS）对103个乳腺结节进行良恶性鉴别诊断，以术后病理为金标准，比较3种诊断方法的诊断效能，并对比分析S-detect应用于CUS（CUS + S-detect）、UE应用于CUS（CUS + UE）以及S-detect联合UE应用于CUS（CUS + S-detect + UE）的诊断效能。
结果： CUS、S-detect及UE 3种诊断方法的灵敏度、特异度、准确度、AUC比较差异均无统计学意义（P ＞ 0.05）。CUS、CUS + S-detect、CUS + UE、CUS + S-detect + UE的灵敏度分别为78.05%、87.80%、75.61%、92.68%，特异度分别为69.35%、82.26%、85.48%、93.55%，准确度分别为72.82%、 84.47%、81.55%、93.20%，ROC曲线下面积（AUC）分别为0.737、0.850、0.805、0.931。与单独CUS比较，CUS + S-detect具有更高的准确度和AUC值，而CUS + UE具有更高的特异度（P ＜ 0.05）。CUS + S-detect + UE联合诊断的准确度和AUC值均高于CUS + S-detect或CUS + UE（P ＜ 0.05）。
结论： CUS + S-detect + UE联合诊断时可实现优势互补，可明显提高乳腺良恶性结节的诊断效能。

Abstract:
Objective To explore the diagnostic value of artificial intelligence S-Detect technology combined with ultrasound elastography (UE) in the differential diagnosis of benign and malignant breast nodules.
Methods A total of 103 breast nodules were subjected to benign-malignant differentiation using S-Detect, UE, and conventional ultrasound (CUS). Postoperative pathology served as the gold standard. The diagnostic efficacy of the three methods was compared, and the diagnostic performance of S-Detect applied to CUS (CUS + S-Detect), UE applied to CUS (CUS + UE), and S-Detect combined with UE applied to CUS (CUS + S-Detect + UE) was analyzed comparatively.
Results There were no statistically significant differences in sensitivity, specificity, accuracy, or AUC among the three diagnostic methods (CUS, S-Detect, and UE) (P ＞ 0.05). The sensitivities of CUS, CUS + S-Detect, CUS + UE, and CUS + S-Detect + UE were 78.05%, 87.80%, 75.61%, and 92.68%, respectively, while the specificities were 69.35%, 82.26%, 85.48%, and 93.55%, the accuracies were 72.82%, 84.47%, 81.55%, and 93.20%, and the AUCs were 0.737, 0.850, 0.805, and 0.931, respectively. Compared to CUS alone, CUS + S-Detect demonstrated higher accuracy and AUC values, while CUS + UE exhibited higher specificity (P ＜ 0.05). The combined diagnostic accuracy and AUC values of CUS + S-Detect + UE were higher than those of CUS + S-Detect or CUS + UE (P ＜ 0.05).
Conclusions When combined, CUS + S-Detect + UE can achieve complementary advantages and significantly improve the diagnostic efficacy for benign and malignant breast nodules.

HTML全文

参考文献(12)

施引文献

资源附件(0)