人工智能超声对乳腺癌早期诊断及预后评估的价值分析

Value analysis of artificial intelligence ultrasound for early diagnosis and prognosis evaluation of breast cancer

目的探讨人工智能超声对乳腺癌早期诊断及预后评估的可行性及价值。方法收集310例疑似早期乳腺癌患者的病历资料,所有患者均进行人工智能超声和彩色多普勒超声检查,以病理学检查结果作为金标准,采用Kappa一致性检验分析人工智能超声和彩色多普勒超声诊断早期乳腺癌与病理结果的一致性。所有确诊患者均进行手术治疗,根据预后复发情况分为复发组(n=10)和未复发组(n=78),比较治疗前后两组患者的超声表现。结果310例疑似乳腺癌患者经过病理学检查,阳性88例,阴性222例。彩色多普勒超声检查诊断乳腺癌的灵敏度、特异度、准确度分别为84.09%、70.72%、74.52%,人工智能超声检查诊断乳腺癌的灵敏度、特异度、准确度分别为86.36%、79.28%、81.29%。彩色多普勒超声和人工智能超声诊断乳腺癌与病理结果的一致性均一般(0.40﹤Kappa值﹤0.75),但人工智能超声诊断乳腺癌与病理结果的一致性更高。治疗后,复发组和未复发组患者超声表现中组织厚度、回声和血流形态比较,差异均有统计学意义(P﹤0.01)。结论人工智能超声可以提高临床超声医师诊断乳腺癌的准确度,同时对预后复发的评估具有较高的可行性。

Objective To investigate the feasibility and value of artificial intelligence ultrasound for early diagnosis and prognosis evaluation of breast cancer.Method The medical records of 310 patients with suspected early breast cancer were collected.All patients underwent artificial intelligence ultrasound and color Doppler ultrasound examination.The pathological examination results were used as the gold standard.The Kappa consistency test was used to analyze the consistency between artificial intelligence ultrasound and color Doppler ultrasound in the diagnosis of early breast cancer and pathological results.All confirmed patients underwent surgical treatment and were divided into recurrence group(n=10)and non-recurrence group(n=78)according to the prognosis.The ultrasound findings of the two groups of patients before and after the treatment were compared.Result A total of 310 patients with suspected breast cancer underwent pathological examination,of which 88 were positive and 222 were negative.The sensitivity,specificity,and accuracy of color Doppler ultrasound in diagnosing breast cancer were 84.09%,70.72%,and 74.52%,respectively.The sensitivity,specificity,and accuracy of artificial intelligence ultrasound in diagnosing breast cancer were 86.36%,79.28%,and 81.29%,respectively.The consistency between color Doppler ultrasound and artificial intelligence ultrasound in the diagnosis of breast cancer and pathological results were general(0.40<Kappa value<0.75),while the consistency between artificial intelligence ultrasound in the diagnosis of breast cancer and pathological results was higher.After the treatment,there were statistically significant differences in tissue thickness,echo,and blood flow morphology between the recurrence group and the non-recurrence group(P<0.01).Conclusion Artificial intelligence ultrasound could improve the accuracy of clinical sonographers in diagnosing breast cancer,and it has high feasibility for the evaluation of prognosis and recurrence at the same time.