The documentation and quantification of seizures is the primary outcome measure of epilepsy therapy including anti epileptic drug treatment, epilepsy surgery, and neurostimulation. There is increasing amount of data suggesting that traditional seizure diaries are highly unreliable and suffer from underreporting with sensitivities of 50% during day and as low as 30% during night. This inaccuracy forms a major issue for the assessment of treatment efficacy .
Recent academic research strongly supports video analysis as an objective measurement tool for motorised epileptic seizures , with evidence that these biomarkers can be used for risk analysis in patients with seizure symptoms . Motion analysis, in particular, has been shown to be an effective technique for detecting the presence of a motor seizure with high sensitivity . With emerging technologies based on deep learning, it is possible to build models not simply based on motion, but that of the human skeleton . By combining traditional signal processing techniques with physical musculoskeletal models, a higher specificity on the type and severity of such seizures can be achieved.
 F. Fürbass et al, Automatic multimodal detection fo long term seizure documentation in epilepsy.
 Ulate-Campos et al, Automated seizure detection systems and their effectiveness for each type of seizure.
 Alexandre et al, Risk factors of postictal generalized EEG suppression in generalized convulsive seizures.
 Cattani et al, Monitoring infants by automatic video processing: A unified approach to motion analysis.
 Wei et al, Convolutional Pose Machines, CVPR 2016.
Clinical validation of sensitivity and specificity of seizure detection at a home environment compared to clinical neurophysiological professionals’ analysis.
Clinical validation of sensitivity and specificity of seizure detection in Helsinki University Hospital compared to traditional video-EEG registering.
Computer vision based analysis of myoclonus, compared to traditional human scoring method.
Analyzing epilepsy seizures’ impact on sleep using computer vision and machine intelligence partially in parallel with ambulatory EEG at a home environment.
Quantitative and objective analysis of seizure and its components using computer vision and machine learning.
Clinical validation of detecting sleep disorders and breathing problems using non-invasive video based movement analysis in Tampere University Hospital sleeping lab (vEEG).
Jukka Peltola, MD, PhD
Chief Physician, Dept Neurology, Tampere University Hospital Professor, Dept Neurology, University of Tampere