Deep-learning for automated markerless tracking of infants general movements

Author:

Abbasi H.ORCID,Mollet S. R.,Williams S. A.,Lim L.,Battin M. R.,Besier T. F.,McMorland A. J. C.

Abstract

AbstractThe presence of abnormal infant General Movements (GMs) is a strong predictor of progressive neurodevelopmental disorders, including cerebral palsy (CP). Automation of the assessment will overcome scalability barriers that limit its delivery to at-risk individuals. Here, we report a robust markerless pose-estimation scheme, based on advanced deep-learning technology, to track infant movements in consumer mobile device video recordings. Two deep neural network models, namely Efficientnet-b6 and resnet-152, were trained on manually annotated data across twelve anatomical locations (3 per limb) in 12 videos from 6 full-term infants (mean age = 17.33 (SD 2.9) wks, 4 male, 2 female), using the DeepLabCut™ framework. K-fold cross-validation indicates the generalization capability of the deep-nets for GM tracking on out-of-domain data with an overall performance of 95.52% (SD 2.43) from the best performing model (Efficientnet-b6) across all infants (performance range: 84.32–99.24% across all anatomical locations). The paper further introduces an automatic, unsupervised strategy for performance evaluation on extensive out-of-domain recordings through a fusion of likelihoods from a Kalman filter and the deep-net. Findings indicate the possibility of establishing an automated GM tracking platform, as a suitable alternative to, or support for, the current observational protocols for early diagnosis of neurodevelopmental disorders in early infancy.

Funder

Friedlanders Foundation

University of Auckland

Publisher

Springer Science and Business Media LLC

Subject

Electrical and Electronic Engineering,Applied Mathematics,Artificial Intelligence,Computational Theory and Mathematics,Computer Networks and Communications,Computer Science Applications,Information Systems

Cited by 6 articles. 订阅此论文施引文献 订阅此论文施引文献,注册后可以免费订阅5篇论文的施引文献,订阅后可以查看论文全部施引文献

1. Performance Comparison of Deep Learning Methods for Lung Nodule Benign and Malignancy Classification;2024 11th International Conference on Computing for Sustainable Global Development (INDIACom);2024-02-28

2. Task Scheduling Symphony: Reinforcement Learning for Fog Computing;2024 11th International Conference on Computing for Sustainable Global Development (INDIACom);2024-02-28

3. Identification and Detection of Leaf Miner, Pests Infestation in Cucurbitaceae Family in Real Time Infield Scenarios using YOLOv5s Object Detection Model;2024 11th International Conference on Computing for Sustainable Global Development (INDIACom);2024-02-28

4. Vehicle Intrusion Classification using Deep Learning and Simulated Sensor Networks;2024 11th International Conference on Computing for Sustainable Global Development (INDIACom);2024-02-28

5. Catalyzing Educational Transformation: A Comprehensive Exploration of High-Performance Computing Integration and Future Trajectories;2024 11th International Conference on Computing for Sustainable Global Development (INDIACom);2024-02-28

同舟云学术

1.学者识别学者识别

2.学术分析学术分析

3.人才评估人才评估

"同舟云学术"是以全球学者为主线,采集、加工和组织学术论文而形成的新型学术文献查询和分析系统,可以对全球学者进行文献检索和人才价值评估。用户可以通过关注某些学科领域的顶尖人物而持续追踪该领域的学科进展和研究前沿。经过近期的数据扩容,当前同舟云学术共收录了国内外主流学术期刊6万余种,收集的期刊论文及会议论文总量共计约1.5亿篇,并以每天添加12000余篇中外论文的速度递增。我们也可以为用户提供个性化、定制化的学者数据。欢迎来电咨询!咨询电话:010-8811{复制后删除}0370

www.globalauthorid.com

TOP

Copyright © 2019-2024 北京同舟云网络信息技术有限公司
京公网安备11010802033243号  京ICP备18003416号-3