HyeonWoo Lee, William Shi, Rashid Al Mukaddim, Elizabeth Brunelle, Abhinav Palisetti, Syed M Imaduddin, Phavalan Rajendram, Diego Incontri, Vasileios-Arsenios Lioutas, Thomas Heldt, Balasundar I Raju
{"title":"经颅多普勒超声检查中基于深度学习的大脑中动脉自动检测。","authors":"HyeonWoo Lee, William Shi, Rashid Al Mukaddim, Elizabeth Brunelle, Abhinav Palisetti, Syed M Imaduddin, Phavalan Rajendram, Diego Incontri, Vasileios-Arsenios Lioutas, Thomas Heldt, Balasundar I Raju","doi":"10.1016/j.ultrasmedbio.2025.05.028","DOIUrl":null,"url":null,"abstract":"<p><strong>Objective: </strong>Transcranial Doppler (TCD) ultrasound has significant clinical value for assessing cerebral hemodynamics, but its reliance on operator expertise limits broader clinical adoption. In this work, we present a lightweight real-time deep learning-based approach capable of automatically identifying the middle cerebral artery (MCA) in TCD Color Doppler images.</p><p><strong>Methods: </strong>Two state-of-the-art object detection models, YOLOv10 and Real-Time Detection Transformers (RT-DETR), were investigated for automated MCA detection in real-time. TCD Color Doppler data (41 subjects; 365 videos; 61,611 frames) were collected from neurologically healthy individuals (n = 31) and stroke patients (n = 10). MCA bounding box annotations were performed by clinical experts on all frames. Model training consisted of pretraining utilizing a large abdominal ultrasound dataset followed by subsequent fine-tuning on acquired TCD data. Detection performance at the instance and frame levels, and inference speed were assessed through four-fold cross-validation. Inter-rater agreement between model and two human expert readers was assessed using distance between bounding boxes and inter-rater variability was quantified using the individual equivalence coefficient (IEC) metric.</p><p><strong>Results: </strong>Both YOLOv10 and RT-DETR models showed comparable frame level accuracy for MCA presence, with F1 scores of 0.884 ± 0.023 and 0.884 ± 0.019 respectively. YOLOv10 outperformed RT-DETR for instance-level localization accuracy (AP: 0.817 vs. 0.780) and had considerably faster inference speed on a desktop CPU (11.6 ms vs. 91.14 ms). Furthermore, YOLOv10 showed an average inference time of 36 ms per frame on a tablet device. The IEC was -1.08 with 95 % confidence interval: [-1.45, -0.19], showing that the AI predictions deviated less from each reader than the readers' annotations deviated from each other.</p><p><strong>Conclusion: </strong>Real-time automated detection of the MCA is feasible and can be implemented on mobile platforms, potentially enabling wider clinical adoption by less-trained operators in point-of-care settings.</p>","PeriodicalId":49399,"journal":{"name":"Ultrasound in Medicine and Biology","volume":" ","pages":""},"PeriodicalIF":2.4000,"publicationDate":"2025-06-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":"{\"title\":\"Deep Learning-Based Automated Detection of the Middle Cerebral Artery in Transcranial Doppler Ultrasound Examinations.\",\"authors\":\"HyeonWoo Lee, William Shi, Rashid Al Mukaddim, Elizabeth Brunelle, Abhinav Palisetti, Syed M Imaduddin, Phavalan Rajendram, Diego Incontri, Vasileios-Arsenios Lioutas, Thomas Heldt, Balasundar I Raju\",\"doi\":\"10.1016/j.ultrasmedbio.2025.05.028\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"<p><strong>Objective: </strong>Transcranial Doppler (TCD) ultrasound has significant clinical value for assessing cerebral hemodynamics, but its reliance on operator expertise limits broader clinical adoption. In this work, we present a lightweight real-time deep learning-based approach capable of automatically identifying the middle cerebral artery (MCA) in TCD Color Doppler images.</p><p><strong>Methods: </strong>Two state-of-the-art object detection models, YOLOv10 and Real-Time Detection Transformers (RT-DETR), were investigated for automated MCA detection in real-time. TCD Color Doppler data (41 subjects; 365 videos; 61,611 frames) were collected from neurologically healthy individuals (n = 31) and stroke patients (n = 10). MCA bounding box annotations were performed by clinical experts on all frames. Model training consisted of pretraining utilizing a large abdominal ultrasound dataset followed by subsequent fine-tuning on acquired TCD data. Detection performance at the instance and frame levels, and inference speed were assessed through four-fold cross-validation. Inter-rater agreement between model and two human expert readers was assessed using distance between bounding boxes and inter-rater variability was quantified using the individual equivalence coefficient (IEC) metric.</p><p><strong>Results: </strong>Both YOLOv10 and RT-DETR models showed comparable frame level accuracy for MCA presence, with F1 scores of 0.884 ± 0.023 and 0.884 ± 0.019 respectively. YOLOv10 outperformed RT-DETR for instance-level localization accuracy (AP: 0.817 vs. 0.780) and had considerably faster inference speed on a desktop CPU (11.6 ms vs. 91.14 ms). Furthermore, YOLOv10 showed an average inference time of 36 ms per frame on a tablet device. The IEC was -1.08 with 95 % confidence interval: [-1.45, -0.19], showing that the AI predictions deviated less from each reader than the readers' annotations deviated from each other.</p><p><strong>Conclusion: </strong>Real-time automated detection of the MCA is feasible and can be implemented on mobile platforms, potentially enabling wider clinical adoption by less-trained operators in point-of-care settings.</p>\",\"PeriodicalId\":49399,\"journal\":{\"name\":\"Ultrasound in Medicine and Biology\",\"volume\":\" \",\"pages\":\"\"},\"PeriodicalIF\":2.4000,\"publicationDate\":\"2025-06-28\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"\",\"citationCount\":\"0\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"Ultrasound in Medicine and Biology\",\"FirstCategoryId\":\"3\",\"ListUrlMain\":\"https://doi.org/10.1016/j.ultrasmedbio.2025.05.028\",\"RegionNum\":3,\"RegionCategory\":\"医学\",\"ArticlePicture\":[],\"TitleCN\":null,\"AbstractTextCN\":null,\"PMCID\":null,\"EPubDate\":\"\",\"PubModel\":\"\",\"JCR\":\"Q2\",\"JCRName\":\"ACOUSTICS\",\"Score\":null,\"Total\":0}","platform":"Semanticscholar","paperid":null,"PeriodicalName":"Ultrasound in Medicine and Biology","FirstCategoryId":"3","ListUrlMain":"https://doi.org/10.1016/j.ultrasmedbio.2025.05.028","RegionNum":3,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q2","JCRName":"ACOUSTICS","Score":null,"Total":0}
引用次数: 0
摘要
目的:经颅多普勒超声(TCD)在评估脑血流动力学方面具有重要的临床价值,但其对操作者专业知识的依赖限制了其在临床的广泛应用。在这项工作中,我们提出了一种基于轻量级实时深度学习的方法,能够自动识别TCD彩色多普勒图像中的大脑中动脉(MCA)。方法:采用两种最先进的目标检测模型YOLOv10和实时检测变压器(RT-DETR)进行实时自动MCA检测。TCD彩色多普勒数据(41例;365个视频;61,611帧)从神经健康个体(n = 31)和中风患者(n = 10)中收集。临床专家对所有帧进行MCA边界框标注。模型训练包括利用大型腹部超声数据集进行预训练,随后对获得的TCD数据进行微调。通过四次交叉验证评估了实例和帧级别的检测性能以及推理速度。使用边界框之间的距离来评估模型与两个人类专家读者之间的评级一致性,并使用个体等效系数(IEC)度量来量化评级间的可变性。结果:YOLOv10和RT-DETR模型对MCA存在的帧级精度相当,F1得分分别为0.884±0.023和0.884±0.019。YOLOv10在实例级定位精度方面优于RT-DETR (AP: 0.817 vs. 0.780),并且在桌面CPU上具有更快的推理速度(11.6 ms vs. 91.14 ms)。此外,YOLOv10在平板设备上的平均推理时间为每帧36毫秒。IEC为-1.08,95%置信区间为[-1.45,-0.19],表明AI预测与每个读者之间的偏差小于读者注释之间的偏差。结论:MCA的实时自动检测是可行的,可以在移动平台上实施,有可能使未经培训的操作员在护理点环境中更广泛地应用于临床。
Deep Learning-Based Automated Detection of the Middle Cerebral Artery in Transcranial Doppler Ultrasound Examinations.
Objective: Transcranial Doppler (TCD) ultrasound has significant clinical value for assessing cerebral hemodynamics, but its reliance on operator expertise limits broader clinical adoption. In this work, we present a lightweight real-time deep learning-based approach capable of automatically identifying the middle cerebral artery (MCA) in TCD Color Doppler images.
Methods: Two state-of-the-art object detection models, YOLOv10 and Real-Time Detection Transformers (RT-DETR), were investigated for automated MCA detection in real-time. TCD Color Doppler data (41 subjects; 365 videos; 61,611 frames) were collected from neurologically healthy individuals (n = 31) and stroke patients (n = 10). MCA bounding box annotations were performed by clinical experts on all frames. Model training consisted of pretraining utilizing a large abdominal ultrasound dataset followed by subsequent fine-tuning on acquired TCD data. Detection performance at the instance and frame levels, and inference speed were assessed through four-fold cross-validation. Inter-rater agreement between model and two human expert readers was assessed using distance between bounding boxes and inter-rater variability was quantified using the individual equivalence coefficient (IEC) metric.
Results: Both YOLOv10 and RT-DETR models showed comparable frame level accuracy for MCA presence, with F1 scores of 0.884 ± 0.023 and 0.884 ± 0.019 respectively. YOLOv10 outperformed RT-DETR for instance-level localization accuracy (AP: 0.817 vs. 0.780) and had considerably faster inference speed on a desktop CPU (11.6 ms vs. 91.14 ms). Furthermore, YOLOv10 showed an average inference time of 36 ms per frame on a tablet device. The IEC was -1.08 with 95 % confidence interval: [-1.45, -0.19], showing that the AI predictions deviated less from each reader than the readers' annotations deviated from each other.
Conclusion: Real-time automated detection of the MCA is feasible and can be implemented on mobile platforms, potentially enabling wider clinical adoption by less-trained operators in point-of-care settings.
期刊介绍:
Ultrasound in Medicine and Biology is the official journal of the World Federation for Ultrasound in Medicine and Biology. The journal publishes original contributions that demonstrate a novel application of an existing ultrasound technology in clinical diagnostic, interventional and therapeutic applications, new and improved clinical techniques, the physics, engineering and technology of ultrasound in medicine and biology, and the interactions between ultrasound and biological systems, including bioeffects. Papers that simply utilize standard diagnostic ultrasound as a measuring tool will be considered out of scope. Extended critical reviews of subjects of contemporary interest in the field are also published, in addition to occasional editorial articles, clinical and technical notes, book reviews, letters to the editor and a calendar of forthcoming meetings. It is the aim of the journal fully to meet the information and publication requirements of the clinicians, scientists, engineers and other professionals who constitute the biomedical ultrasonic community.