Multi-modal Ultrasound Combined with Artificial Intelligence S-Detect Technology for Diagnosis of BI-RADS 4 Types of Breast Nodules
Objective:To analyze the diagnostic value of multimodal ultrasound combined with artifi-cial intelligence S-Detect technology in breast Image Reporting Data System(BI-RADS)for 4 types of breast nodules.Methods:Sixty-two patients with breast lesions admitted to our hospital from September 2020 to March 2024 were retrospectively collected,and the clinical data of all patients were statistically analyzed.The diagnostic results of different diagnostic methods were compared,and the diagnostic effi-ciency of multi-modal ultrasound combined with artificial intelligence S-Detect technology for BI-RADS type 4 breast nodules was analyzed with pathological diagnosis as the gold standard.Results:The features of contra-ultrasound of malignant breast nodules are:high enhancement,centripetal enhancement,un-even internal echo,irregular shape,unclear edge,presence of peripheral radial vessels,and increased le-sion volume after enhancement.There were 76 breast nodules in 62 patients in this group.Pathological re-sults showed that 42 nodules were malignant and 34 nodules were benign.34 malignant cases were diag-nosed by conventional ultrasound,the sensitivity was 71.43％,the specificity was 88.24％and the accu-racy was 78.95％.The artificial intelligence S-Detect technology diagnosed 39 malignancies,the sensitiv-ity was 76.19％,the specificity was 79.41％,and the accuracy was 77.63％.The sensitivity,specificity and accuracy of 44 malignancies were 85.71％,76.47％and 81.58％respectively.The multi-modal ultra-sound combined with artificial intelligence S-Detect technology diagnosed 45 malignancies,with a sensi-tivity of 95.24％,specificity of 85.29％and accuracy of 90.79％.The pathological results were taken as the gold standard,and the diagnosis results of multi-modal ultrasound combined with artificial intelli-gence S-Detect technology were highly consistent with the pathological results,with the highest consis-tency.The Kappa value was 0.812,the sensitivity was 95.24％,the specificity was 85.29％and the accu-racy was 90.79％.Comparative analysis of ROC curve showed that the diagnostic efficiency of multi-mode ultrasound combined with artificial intelligence S-Detect technology was statistically significant compared with that of conventional ultrasound alone,artificial intelligence S-Detect technology and multi-mode ultrasound(Z=0.275/2.603/2.083,P=0.023/0.009/0.037).Conclusion:Compared with single ultrasound,multi-modal ultrasound and artificial intelligence S-Detect technology,multi-modal ultra-sound combined with artificial intelligence S-Detect technology has higher diagnostic efficiency in the di-agnosis of BI RADS type 4 breast nodules,which is highly consistent with pathological results.It can be popularized and applied.
Breast nodulesBI-RADS classificationMultimode ultrasoundArtificial intelligence S-Detect technology
NETL
NSTL
万方数据
