人工智能在早孕期胎儿颅脑超声切面质量评价中的应用
Application of Artificial Intelligence for Quality Evaluation of Fetal Cranial Ultrasound Planes in Early Pregnancy
郭旭 1林艺榕 2余卫峰 1吴秀明 1柳培忠 3刘中华1
作者信息
- 1. 福建医科大学附属泉州第一医院超声科 福建省泉州市,362000
- 2. 华侨大学医学院 福建省泉州市,362000
- 3. 华侨大学工学院 福建省泉州市,362000
- 折叠
摘要
目的 探讨基于YOLOv5-s的人工智能(AI)模型在早孕期胎儿颅脑超声切面质量评价中的应用价值.方法 以妊娠11～13+6周正常胎儿超声检查图像为研究对象,其中模型构建集(3 094张颅脑超声图像)用于AI模型的训练和验证,实验集184胎早孕期胎儿常规超声检查图像(7 930张图像)用于进行早孕期颅脑切面质量评价.分析AI模型与2位专家级超声医师对实验集颅脑切面质控结果的一致性.结果 AI模型对验证集各切面识别准确度均在96.00％以上,与专家级超声医师分类具有极强一致性.AI模型与专家级超声医师-2对实验集经侧脑室横切面质控结果具有中度一致性,其余质控结果AI模型与2位专家级超声医师两两比较均具有高度或极强一致性.结论 AI模型识别早孕期颅脑超声切面具有较高的准确度,可用于进行超声图像质量评价以及培训考核.
Abstract
Objective To explore the value of YOLOv5-s based artificial intelligence(AI)model in the quality evaluation of fetal cranial ultrasound planes in early pregnancy.Methods Normal fetal ultrasound images at 11～13+6 weeks of gestation were used for the study,where the model construction set(3 094 cranial ultrasound images)was used for the training and validation of the AI model,and 184 routine ultrasound images of the fetus in early pregnancy in the experimental set(7 930 images)was used to perform the cranial planes'quality evaluation.The consistency of the quality control results of cranial brain planes in the experimental set by AI model and two ultrasound experts were analyzed.Results The AI model had more than 96.00％recognition accuracy for all planes in the validation set,showing very strong consistency with ultrasound experts'classifications.There was moderate consistency between the quality control results for the trans-lateral ventricular planes in the experimental set by the AI model and ultrasound expert-2,and the rest of the two-by-two comparison results of quality control results showed a high or very strong con-sistency.Conclusions The application of the AI model to identify cranial ultrasound images of early pregnancy has high accuracy and can be used to perform ultrasound image quality evaluation as well as training examinations.
关键词
人工智能/超声检查/早孕期/胎儿/颅脑超声切面/质量控制Key words
Artificial intelligence/Ultrasound/Early pregnancy/Fetus/Cranial ultrasound plane/Quality con-trol引用本文复制引用
出版年
2025
NETL
NSTL
万方数据