| 罗丹丹,郭文佳,曾晴,余翔,付倩,秦越,曾绮汶,欧阳淑媛,李胜利,袁鹰,文华轩.人工智能自动获取胎儿超声心动图标准切面[J].中国医学影像技术,2026,42(6):941~945 |
| 人工智能自动获取胎儿超声心动图标准切面 |
| Artificial intelligence for automatically acquiring standard sections of fetal echocardiography |
| 投稿时间：2026-02-12 修订日期：2026-06-13 |
| DOI：10.13929/j.issn.1003-3289.2026.06.031 |
| 中文关键词: 胎儿心脏 超声检查,产前 人工智能 |
| 英文关键词:fetal heart ultrasonography, prenatal artificial intelligence |
| 基金项目:深圳市科技计划(JCYJ20230807120304009)。 |
| 作者 | 单位 | E-mail | | 罗丹丹 | 南方医科大学妇女儿童医学中心深圳市妇幼保健院超声科, 广东 深圳 518028 | | | 郭文佳 | 南方医科大学妇女儿童医学中心深圳市妇幼保健院超声科, 广东 深圳 518028 | | | 曾晴 | 南方医科大学妇女儿童医学中心深圳市妇幼保健院超声科, 广东 深圳 518028 | | | 余翔 | 南方医科大学妇女儿童医学中心深圳市妇幼保健院超声科, 广东 深圳 518028 | | | 付倩 | 南方医科大学妇女儿童医学中心深圳市妇幼保健院超声科, 广东 深圳 518028 | | | 秦越 | 南方医科大学妇女儿童医学中心深圳市妇幼保健院超声科, 广东 深圳 518028 | | | 曾绮汶 | 南方医科大学妇女儿童医学中心深圳市妇幼保健院超声科, 广东 深圳 518028 | | | 欧阳淑媛 | 南方医科大学妇女儿童医学中心深圳市妇幼保健院超声科, 广东 深圳 518028 | | | 李胜利 | 南方医科大学妇女儿童医学中心深圳市妇幼保健院超声科, 广东 深圳 518028 | | | 袁鹰 | 南方医科大学妇女儿童医学中心深圳市妇幼保健院超声科, 广东 深圳 518028 | | | 文华轩 | 南方医科大学妇女儿童医学中心深圳市妇幼保健院超声科, 广东 深圳 518028 | whxwell@126.com |
|
| 摘要点击次数: 153 |
| 全文下载次数: 32 |
| 中文摘要: |
| 目的 观察人工智能(AI)用于自动获取胎儿超声心动图标准切面的价值。方法 前瞻性招募拟接受超声产前筛查和胎儿心动图检查的151名健康单胎孕妇,分别采用人工方法与AI技术获取胎儿心脏11个标准切面灰阶及CDFI图像;比较2种方法获取图像成功率及图像标准率。随机选取40胎,比较2种方法获取切面所需时间。结果 AI获取各标准切面的成功率98.68%~100%,人工为90.07%~100%;AI获取心底短轴切面和三血管切面二维灰阶及CDFI图像,以及动脉导管弓长轴切面和上下腔静脉长轴切面二维灰阶图像成功率均高于人工(P均＜0.05)。AI获取切面图像标准率为94.67%~100%,人工为97.79%~100%;2种方法获取各切面图像标准率差异均无统计学意义(P均＞0.05)。AI获取标准切面平均用时(205.6±113.2)s,短于人工方法的(317.7±113.7)s(P＜0.05)。结论 AI自动获取胎儿心脏标准切面成功率较高且用时较少。 |
| 英文摘要: |
| Objective To explore the value of artificial intelligence (AI) technique for automatically acquiring standard sections of fetal echocardiography. Methods A total of 151 healthy singleton pregnancies scheduled for prenatal ultrasound screening and fetal echocardiography were prospectively recruited. Both grayscale and CDFI images on 11 standard fetal cardiac sections were acquired manually and with AI automated method, respectively. The success rate of section acquisition and standard rate of the acquired sections were compared between the two methods. Then 40 fetuses were randomly selected, the time spent on image acquisition was compared between the two methods. Results The success rate of section acquision of AI method was 98.68%—100%, of manual method was 90.07%—100%. The success rate of AI was higher than manual methods for acquiring two-dimensional (2D) grayscale and CDFI images of the short axis sections and three-vessel section, also for 2D grayscale images of long-axis section of the ductal arch and long axis section of superior and inferior vena cava (all P＜0.05). The standard rate of AI and manual methods was 94.67%—100% and 97.79%—100%, respectively, without significant difference for any section (all P＞0.05). The average time spent of AI for acquiring standard sections was shorter than of manual method ([205.6±113.2] s vs. [317.7±113.7] s, P＜0.05). Conclusion AI achieved high success rate in automatically acquiring standard sections of fetal echocardiography with less time. |
| 查看全文 查看/发表评论 下载PDF阅读器 |
|
|
|
