Search

The authors propose a glove with pressure sensitive fingertips and muscle tension detection of the thumb to aid in the physiotherapy of infants, children and adults according to Vojta. The therapy has to be carried out at home by the parents of children with cerebral palsy, or other movement disturbances up to 4 times per day. Often, parents struggle to carry out the practices because of insecurities regarding the therapy’s proper application. The presented glove is tailored towards the specific needs of Vojta therapy, e.g. flexibility and small sensor size, and can help parents gather objective data on their therapy application at home. It has been tested in multiple therapy sessions on adult probands and seems to be a suitable tool to support parents and therapists in training alike. Additionally, the results of the accompanied therapy sessions provide arguments against what the critics have often cited about the painfulness of the therapy.

Striving for more robust and natural control of multi-fingered hand prostheses, we are studying electrical impedance tomography (EIT) as a method to monitor residual muscle activations. Previous work has shown promising results for hand gesture recognition, but also lacks generalization across multiple sessions and users. Thus, the present paper aims for a detailed analysis of an existing EIT dataset acquired with a 16-electrode wrist band as a prerequisite for further improvements of machine learning results on this type of signal. The performed t-SNE analysis confirms a much stronger inter-session and inter-user variance compared to the expected in-class variance. Additionally, we observe a strong drift of signals within a session. To handle these challenging problems, we propose new machine learning architectures based on deep learning, which allow to separate undesired from desired variation and thus significantly improve the classification accuracy. With these new architectures we increased cross-session classification accuracy on 12 gestures from 19.55 to 30.45%. Based on a fundamental data analysis we developed three calibration methods and thus were able to further increase cross-session classification accuracy to 39.01, 55.37, and 56.34%, respectively.