AI超声智能辅助诊断系统对甲状腺结节诊断价值的初步探讨

Preliminary study of diagnostic value of AI intelligent assistant diagnostic system for thyroid nodules

目的:利用人工智能(AI)超声智能辅助诊断系统评估甲状腺结节的良恶性,并与不同年资超声医师的诊断效能进行对照分析,探讨AI在甲状腺结节诊断中的应用价值。方法:回顾性分析有明确病理结果的甲状腺结节超声图像,由3名超声医师独立分析,按照Kwak版TI-RADS标准进行分类。同时利用AI智能辅助诊断系统进行TI-RADS分类并给出良恶性概率值。结果:本研究共分析了176例195个甲状腺结节,包括恶性结节86个,良性结节109个。以TI-RADS 4a或4b以上考虑恶性,AI诊断的敏感性、特异性、准确性及ROC曲线下面积(0.955)均最高,高年资医师次之,而低年资医师和社区医师的诊断效能相对较低。当AI恶性概率的截断值为0.48,Youden指数为0.812时,其敏感性和特异性分别达到89.5%,91.7%,具有最佳诊断效能。结论:AI智能辅助诊断系统对甲状腺结节具有较高的诊断价值,诊断效能明显优于低年资超声医师和社区医师,具有广阔的临床应用前景。

Objective:To evaluate the benign and malignant of thyroid nodules using artifitial intelligence(AI)ultrasound auxiliary diagnosis system,and compare it with the diagnostic efficiency of ultrasound doctors with different seniority,so as to explore the value of AI in the diagnosis of thyroid nodules.Methods:The ultrasound images of thyroid nodules with definite pathological results were retrospectively analyzed by three ultrasound doctors independently,and classified according to the Kwak TI-RADS standard.At the same time,AI intelligent auxiliary diagnosis system was used to classify and the benign and malignant probability values were given.Results:A total of 176 patients with 195 thyroid nodules were analyzed,including 86 malignant nodules and 109 benign nodules.When TI-RADS 4 a or 4 b was considered as malignant,the sensitivity,specificity and accuracy of AI diagnosis were the highest,and the area under ROC curve was the highest(0.955).That is to say,the diagnostic efficiency of AI was the highest,followed by senior doctors,while the diagnostic efficiency of junior doctors and community doctors was relatively low.When the cutoff value of malignant probability was 0.48 and Youden index was 0.812,the sensitivity and specificity were 89.5% and 91.7%respectively,which had the best diagnostic efficiency.Conclusion:AI intelligent auxiliary diagnosis system has high diagnostic value for thyroid nodules,and its diagnostic efficiency is significantly better than that of junior ultrasound doctors and community doctors,which has broad clinical application prospects.