Prof. Dr.-Ing. Bert Arnrich

Visualization and Quantification Approach of IMU-Derived Stroke Gait Analysis

Master's Thesis

Lennard Ekrod, Supervisor: Lin Zhou

Stroke is a leading cause of disability and long-term health issues. One of the challenges in stroke rehabilitation is the assessment of gait or the way a person walks. In recent years, there has been a growing interest in using inertial measurement units (IMUs) to track and quantify gait data in stroke rehabilitation. IMUs are small, lightweight devices that can be attached to the body to measure movement. There has been a lot of research done to classify stroke compared to regular gait and a growing interest in the classification of stroke gait within a stroke population can be seen in the literature. This thesis questions the clinical relevance of many of these approaches and tries to implement a method with higher clinical relevance. The new method grants important insights into the gait ability of stroke subjects during their rehabilitation process. To achieve this, the study explores the use of visualization and quantification of IMU- derived gait data measured at two-time points in the early rehabilitation stage of stroke patients. The aim is to provide a better understanding of the changes in gait patterns and how they can be tracked and quantified using IMUs. 10 stroke subjects with a FAC severity score ranging from 1-5 were measured on two separate visits with 7-8 days between the two measurements. The gait parameters were extracted using a pipeline introduced by previous work and included spatiotemporal, symmetry, and variation parameters.

The changes in gait parameters between the two visits were visualized using radar plots. To further investigate trends within the population, boxplots displaying the range of values for each subject were also included. The radar plots show a similar gait pattern for subjects with gait ability increases from the first to the second visits and a reversed pattern for subjects with decreases in gait ability between the two visits. They additionally show the individual increases in all three gait parameter groups for each subject and possibly allow an interpretation of gait rehabilitation techniques on short and long-term outcomes. This thesis shows the potential a visualization approach can have and how this might be complementary to classification approaches in the current literature, to further raise the clinical relevance of stroke gait analysis. The method suggested in this thesis, allows the physicians and therapists to engage with the patient and adapt as well as personalize the rehabilitation approach based on the short and long-term changes visualized by the radar plots.