Matthias Mayer

Advisors
Marlies Nitschke (M.Sc.), Dr. Jörg Miehling (KTmfk), Prof. Dr. Sandro Wartzack (KTmfk), Prof. Dr. Jochen Klucken (UK Erlangen), Prof. Dr. Marc Stamminger (LGDV), Prof. Dr. Björn Eskofier, Prof. Dr. Anne Koelewijn

Duration
11/2020 – 07/2021

Abstract

Mobile biosensor systems have been developed to diagnose of Parkinson’s disease (PD) and its severity, since these systems allow for an assessment of gait in the home environment over a longer time period [1]. However, these systems, e.g. the mobile GaitLab system, record only inertial sensors signals from accelerometers and gyroscopes, which are used to determine spatio-temporal variables, such as stride time. Instead, an observational or biomechanical gait analysis could provide much more information about the patient’s gait and thus health state to the clinician.

Recently, we have reconstructed gait in simulation using data from seven inertial sensors attached to the body [2]. Furthermore, we have created a method to visualize these simulations using a human surface model [3]. Now, we would like to employ these two approaches to provide clinicians with detailed gait information based on data from wearable sensor systems, and to allow them to observe the reconstructed gait using the visualization approach.

This thesis aims to design a pipeline that can be used to convert inertial sensor data from the mobile GaitLab system to biomechanical gait simulations, to investigate which biomechanical parameters are different between healthy people and patients with PD, and to investigate if clinicians can identify PD gait based on visualizations of the simulation. To do so, the gait reconstruction method should be extended to handle the sparse sensor setup of the mobile GaitLab system. Furthermore, clustering will be employed to investigate how internal body parameters
are affected by PD. Finally, clinicians will be asked to identify PD gait from gait visualizations.

References

1. Klucken, Jochen et al.: Unbiased and Mobile Gait Analysis Detects Motor Impairment in Parkinson’s Disease. PLOS One, 8(2): e56956, 2013.
2. Dorschky Eva et al.: Estimation of Gait Kinematics and Kinetics from Inertial Sensor Data Using Optimal Control of Musculoskeletal Models. Journal of Biomechanics, 95(11):109278, 2019.
3. Schleicher, Robert, et al. BASH: Biomechanical Animated Skinned Human for Visualization of Kinematics and Muscle Activity. Proceedings of the 16th International Joint Conference on Computer Vision, Imaging and Computer Graphics Theory and Applications – Volume 1: GRAPP 2021.