| 尹炜,李兰涛,周峰峰,冯懿俐,张梦诗,李晓光,尹红霞.基于人工智能分割超高分辨力CT中的内耳半规管并自动测量其形态学参数[J].中国医学影像技术,2026,42(3):425~428 |
| 基于人工智能分割超高分辨力CT中的内耳半规管并自动测量其形态学参数 |
| Segmentation of semicircular canals in ultra-high resolution CT and automatic measurement of morphological parameters based on artificial intelligence |
| 投稿时间：2025-08-19 修订日期：2025-12-11 |
| DOI：10.13929/j.issn.1003-3289.2026.03.021 |
| 中文关键词: 半规管 体层摄影术,X线计算机 人工智能 图像处理,计算机辅助 电子数据处理 |
| 英文关键词:semicircular canals tomography,X-ray computed artificial intelligence image processing,computer-assisted electronic data processing |
| 基金项目:北京市科技计划(Z241100009024020)、国家自然科学基金(62371316)、北京市自然科学基金(7252281)。 |
|
| 摘要点击次数: 92 |
| 全文下载次数: 54 |
| 中文摘要: |
| 目的 观察人工智能(AI)用于分割超高分辨力CT(U-HRCT)图像中的内耳半规管并自动测量其形态学参数的价值。方法 回顾性收集100名(男、女各半)接受U-HRCT检查且左侧内耳结构正常的受试者,采用自主研发的AI模型分割并自动测量左耳前、外、后半规管长度及其两两之间的夹角,并与人工手动测量结果进行对比。结果 各半规管长度及其间夹角的自动测值与手动测值差异均无统计学意义(P均＞0.05)。男性各半规管长度的自动与手动测值均大于女性(P均＜0.05),而不同性别之间左耳半规管间夹角差异均无统计学意义(P均＞0.05)。结论 基于AI能高效、准确地分割U-HRCT中的半规管并自动测其形态学参数;男性半规管长度大于女性。 |
| 英文摘要: |
| Objective To observe the value of artificial intelligence (AI) for segmenting semicircular canals in the inner ear in ultra-high resolution CT (U-HRCT) images and automatically measuring their morphological parameters. Methods A total of 100 subjects with normal left inner ear who underwent U-HRCT examinations were retrospectively enrolled, including 50 males and 50 females. An in-house developed AI segmentation model was used to segment the anterior, outer and posterior semicircular canals of the left ear in U-HRCT images and automaticaly measure the lengths, as well as the spatial angles between each two semicircular canals, and the results were compared with those obtained manually. Results No significant difference of the length nor angles between each two semicircular canals was found between automatic and manual measurements (all P＞0.05). The length of semicircular canals in males obtained automatically and manually were all larger than in females (all P＜0.05), whereas no significant difference of the angles was detected between males and females (all P＞0.05). Conclusion AI could be used to efficiently and accurately segment semicircular canals in U-HRCT and automatically measure their morphological parameters. The length of semicircular canals in males were larger than in females. |
| 查看全文 查看/发表评论 下载PDF阅读器 |
|
|
|
