Dennis Anderson
BETH ISRAEL DEACONESS MEDICAL CENTER/HARVARD MEDICAL SCHOOL
Title: Spine motion and loading from smartphone video – implementing a full spine model in OpenCap
Abstract:
Spine conditions are common, costly, and intimately connected with spine motion and loading. Objective assessments of spine motion and loading in patients could be invaluable to clinicians for diagnostic purposes, as clinical outcome measures, and for developing personalized rehabilitation plans. The goal of this study is to develop a platform for low-cost video-based assessment of spine motion and loading by integrating our advanced musculoskeletal model of the thoracolumbar spine with OpenCap, a platform for cloud-based motion capture and modeling using smartphone cameras. We will first incorporate our validated thoracolumbar musculoskeletal model into the working pipeline of OpenCap. Secondly, we aim to develop a new machine learning algorithm that can estimate the spine marker positions in OpenCap’s pipeline, as these are lacking in current video analysis algorithms. Using our model, spine kinematics and loads during selected tasks will be estimated via OpenCap, and compared with standard marker-based motion analysis to assess the accuracy. The results are expected to demonstrate the potential for low-cost video-based assessments of spine motion and loading, laying the groundwork for additional research efforts in clinical populations, and ultimately translation into clinical practice in rehabilitation.
Kwadwo Appiah-Kubi
CLARKSON UNIVERSITY
Title: Effects of concurrent vestibular activation and postural training on postural control using virtual reality
Collaborators:
- Masudul Imtiaz, PhD. (Clarkson University)
- Emmanuel Asante-Asamani, PhD (Clarkson University)
Abstract:
Vestibular rehabilitation is often effective in resolving vestibular symptoms and improving balance through vestibular recalibration, habituation, sensory reweighting, and postural exercises. However, some groups of patients do not respond to the protocol employed for reasons unknown or unspecified in the study. We showed in previous work that our concurrent vestibular headshake and weight shift training (Concurrent HS-WST) can decrease eye movement variability, improve gaze stabilization, influence sensory weighting, and promote balance using the expensive NeuroCom® Balance Master. The purpose of this pilot study is to determine whether the Concurrent HS-WST delivered by a low-cost, portable, and accessible virtual reality (VR) headset can reweight sensory inputs and modify vestibulo-ocular and balance responses. In a crossover design, 24 healthy participants (55-80 yrs) will be recruited and randomized into two groups. Group one will receive the training intervention for seven days followed by a six-day washout followed by a seven-day period of no-training. Group two will follow the reverse sequence. During the training periods, participants will don an Oculus VR headset and perform horizontal rhythmic headshake activities with simultaneous weight shift training for 25 mins/session. Six pre- and re-assessments of vestibulo-ocular reflex (VOR) gains, postural balance variables and electro-oculography signals will be performed. A repeated measures ANOVA will be used to analyze data. Evidence supporting the use of VR in the Concurrent HS-WST would include significantly decreased VOR gains and eye movement variability and reweighted somatosensory ratio indicative of improved postural behavior. This project will lay the foundation for studies using neurologic patients, such as those with stroke, enabling future development of VR vestibular rehabilitation protocols.
Rachel Hawe
UNIVERSITY OF MINNESOTA
Title: Video-based analysis of bilateral arm use in children with hemiparetic cerebral palsy
Collaborators:
- Stephen Guy, PhD (University of Minnesota)
Abstract:
Children with hemiparetic cerebral palsy (HCP) have motor impairments primarily on one side of their body. A common goal is to increase the use of the more affected arm, especially for tasks typically requiring bilateral arm use. However, few assessment tools are available to assess bilateral arm use. We have developed a video-based analysis pipeline to be able to visualize and quantify bilateral arm use during natural movements. This pipeline uses pose estimation to produce summative figures depicting bilateral arm use and calculate metrics. In this project, we aim to 1) assess the performance of our analysis pipeline in children with HCP, and 2) determine if our analysis can demonstrate differences between children with HCP and typically developing children, and distinguish between different severity levels. Successful implementation of our pipeline in children with HCP will facilitate research and clinical practice in children with HCP by making measures of bilateral arm use more feasible and accessible.
Brooke Odle
HOPE COLLEGE
Title: Classifying Patient-handling techniques to reduce risk of musculoskeletal injury in nursing students
Collaborators:
- Omofolakunmi Olagbemi, PhD (Hope College)
Abstract:
Low back injury risk in nurses is associated with the performance of manual patient-handling tasks. Task performance differs in nursing students and professional nurses despite nursing program curricula incorporating patient handling training. In this study, we will investigate multi-joint coordination of nursing students performing tasks. Participants will include freshmen and sophomores (no clinical skills experience), and juniors and seniors (clinical skills and potentially nursing practicum experience). OpenCap will be used to capture three-dimensional positional data and compute trunk, hip, and knee joint angles during the following tasks: (i) lift and reposition a patient from supine to seated position, (ii) turn a patient over on their side, (iii) lift a patient’s leg, (iv) lift a patient from a wheelchair, (v) slide a patient to the head of the bed using sheets, and (vi) place a sling under a patient. Time spent in particular postures and task completion time will be recorded and compared across each group. Manikins weighing 44, 66, and 110 pounds will serve as the patients. Data collected and patient weight will be inputs to machine learning classifiers that distinguish posture (good, poor, or neutral) for each corresponding task. Successful completion of this project will enhance knowledge about the influence of posture and patient weight on task performance in nursing students, and potentially pave the way to develop a tool that can provide real time feedback on postures adopted by nurses and caregivers during tasks. This work will also provide a foundation for exploring training methods to mitigate risk of injury in nursing students and ultimately nurses in the workforce.