Replication Data for: Automated classification of dystonia and choreoathetosis in dyskinetic cerebral palsy during a lower extremity task: a pilot study on a retrospective video dataset
收藏资源简介：
This RDR repository contains the data and code used in the study ‘Automated classification of dystonia in dyskinetic cerebral palsy within a lower extremity task using markerless motion tracking and time series classification: a pilot study on retrospective videos.’ The study had three main steps: 1. Kinematic Extraction with DeepLabCut 2.3 from videos [1, 2] 2. Post-processing of extracted X,Y coordinates 3. Time-Series Classification with HIVE-COTE 2.0 [3]. Users can begin with our preprocessing script or proceed directly to time series classification with the data provided. Our raw data consisted of 66 videos from 33 participants (29 with dyskinetic cerebral palsy / four typically developing, age 7-23, 13 females / 20 males). The participants performed an item from the Dyskinesia Impairment Scale (DIS) - the heel-toe tapping task with the right lower leg (task 11) and the left lower leg (task 12) [4]. The original videos are NOT included in the RDR repro due to privacy reasons. References: 1. Mathis, A., et al., DeepLabCut: markerless pose estimation of user-defined body parts with deep learning. Nat Neurosci, 2018. 21(9): p. 1281-1289. 2. github.com/DeepLabCut/DeepLabCut 3. Middlehurst M, Large J, Flynn M, Lines J, Bostrom A, Bagnall A. HIVE-COTE 2.0: a new meta ensemble for time series classification. Mach Learn . 2021 Dec 1; 110(11–12):3211–43 4. Monbaliu E, Ortibus E, de Cat J, Dan B, Heyrman L, Prinzie P, et al. The dyskinesia Impairment Scale: a new instrument to measure dystonia and choreoathetosis in dyskinetic cerebral palsy. Dev Med Child Neurol. 2012;54:278–83.
本RDR仓库包含了某项研究的相关数据与代码，该研究标题为《基于无标记运动追踪（markerless motion tracking）与时序分类（time series classification）的运动障碍性脑瘫患者下肢任务中肌张力障碍自动分类：一项回顾性视频先导研究》。该研究主要包含三个核心步骤： 1. 利用DeepLabCut 2.3（DeepLabCut 2.3）从视频[1, 2]中提取运动学数据； 2. 对提取得到的X、Y坐标进行后处理； 3. 采用HIVE-COTE 2.0（HIVE-COTE 2.0）[3]完成时序分类。 用户既可通过我们提供的预处理脚本启动分析流程，也可直接使用已提供的数据开展时序分类任务。 本研究的原始数据来自33名受试者的66段视频：其中29名为运动障碍性脑瘫患者，4名为发育正常者，年龄跨度为7至23岁，女性13名、男性20名。受试者完成了运动障碍损伤量表（Dyskinesia Impairment Scale, DIS）中的两项任务：右侧小腿跟趾叩击任务（任务11）与左侧小腿跟趾叩击任务（任务12）[4]。 出于隐私保护原因，原始视频未包含在本RDR仓库中。 参考文献： 1. Mathis A, 等. DeepLabCut：基于深度学习的用户自定义身体部位无标记姿态估计. 自然神经科学, 2018, 21(9): 1281-1289. 2. github.com/DeepLabCut/DeepLabCut 3. Middlehurst M, Large J, Flynn M, Lines J, Bostrom A, Bagnall A. HIVE-COTE 2.0：一种新型时序分类元集成模型. 机器学习, 2021年12月1日, 110(11–12): 3211–3243. 4. Monbaliu E, Ortibus E, de Cat J, Dan B, Heyrman L, Prinzie P, 等. 运动障碍损伤量表：一种用于评估运动障碍性脑瘫患者肌张力障碍与舞蹈手足徐动症的新工具. 发育医学与儿童神经病学, 2012;54:278–283.
