Ambulatory Movement and Activity Analysis for Patients with Neurological Disorders Based on Wearable Inertial Sensors
Final Report Abstract
Automatic detection of pathological movement patterns caused by neurological disorders is a highly demanded technology with many areas of application in clinical as well as in home care environments. The scientific contributions made during this project were threefold. Firstly, state-of-the-art machine learning models were trained and evaluated for the identification of epileptic seizure events in inertial sensor data. Secondly, the approach was extended to non-invasive monitoring methods, more specifically to range- and infrared-video. Both video modalities were combined in order to maximise the detection rate. Thirdly, a new model was developed that relies on recurrent as well as convolutional neural networks in order to estimate the 3D-pose of a patient in the hospital bed based on range-video data. This model was trained to be robust against blanket occlusion, which greatly improves its usability in a realistic hospital monitoring environment. Possible applications of the research that was conducted range from alerts that can be triggered during seizure events to diagnostically valuable automated seizure counts of nightly seizures that patients are mostly unaware of. Furthermore, a pose estimation method for bedridden patients can be used to support quantitative analysis of movement disorders and could give cues about the status of comatose patients in intensive care units.
Publications
-
“3D Motion Analysis of Epilepsy Patients for Quantitative Comparison of Motor Seizures”. 88. Kongress der Deutschen Gesellschaft für Neurologie, 2015
F. Achilles, L. Casas, A.M. Loesch, E. Hartl, N. Navab, S. Noachtar
-
“Deep convolutional neural networks for automatic identification of epileptic seizures in infrared and depth images”. Journal of the Neurological Sciences 357 (2015): e436
F. Achilles, V. Belagiannis, F. Tombari, A.M. Loesch, J.P.S. Cunha, N. Navab, S. Noachtar
-
“Kombiniertes 3D- und Infrarot-System zur Erkennung epileptischer Anfälle im Schlaf”. Somnologie (2015) 19(Suppl 2): 47
F. Achilles, F. Tombari, V. Belagiannis, A. M. Loesch, J. P. Cunha, N. Navab, S. Noachtar
-
“Convolutional neural networks for real-time epileptic seizure detection”. Computer Methods in Biomechanics and Biomedical Engineering: Imaging & Visualization, July 2016, pp. 1-6
F. Achilles, F. Tombari, V. Belagiannis, A.M. Loesch, S. Noachtar, N. Navab
-
“Patient MoCap: Human Pose Estimation under Blanket Occlusion for Hospital Monitoring Applications”. Medical Image Computing and Computer Assisted Interventions, MICCAI 2016
F. Achilles, A.E. Ichim, H. Coskun, F. Tombari, S. Noachtar, N. Navab
-
“Uncovering epileptic seizures - A feasibility study for the semiological analysis of hidden patient motion during epileptic seizures”. Clinical Neurophysiology 127(9), August 2016, p. e289, Elsevier
F. Achilles, H.M.P. Choupina, A.M. Loesch, J.P.S. Cunha, J. Remi, C. Vollmar, F. Tombari, N. Navab, S. Noachtar
