AI技术在乳腺原位癌Ki67染色切片定量检测中的应用价值

目的探讨AI技术在乳腺原位癌Ki67染色切片定量检测中的应用价值。方法乳腺原位癌110例作为标准样本,通过AI系统辅助检测和专业病理医师人工阅片进行比较,统计两种方法的敏感性、特异性等指标。结果AI辅助检测结果和专业病理医师人工阅片结果显示:Ki67>14%43例,AI阅片和人工阅片阳性比例差异有统计学意义(P<0.05),AI阅片有更高的敏感性;Ki67≤14%67例,AI阅片和人工阅片阳性比例差异无统计学意义(P>0.05)。结论AI独立使用于乳腺原位癌Ki67的筛查尚不成熟。但借用人工智能的识别和深度学习能力,有利于对Ki67的定量检测进行量化,让医师审片更加精准高效,使得检测结果更加精准,为后续对乳腺原位癌的治疗方案提供客观的病理诊断依据。

Objective To explore the application of AI technology in the quantitative detection of Breast Carcinoma in situ with Ki67 staining.Methods Using 110 cases of breast cancer in situ from Deqing County,Zhejiang People's Hospital as a standard sample,a comparative study was conducted by using AI system to assist detection and manual screening by professional pathologists.Results The results of AI assistant examination were compared with the results of manual film-reading by professional pathologists.The results showed that 43 cases with Ki67>14%had the positive rate of AI filmreading and manual film-reading(P<0.05).AI scans are more sensitive.67 cases with Ki67≤14%;the positive rate of AI and manual film reading(P>0.05).Conclusion The application of AI in screening Ki67 breast carcinoma in situ is not mature.But the use of artificial intelligence for recognition and deep learning,is conducive to the quantitative detection of Ki67,which aids the doctors for a more accurate and efficient screening,and the test results more accurate,to provide objective pathological diagnosis basis for the follow-up treatment of breast carcinoma in situ.