Abstract
Epilepsy is one of the most common neurological disorders, affecting up to 1% of the World population. Patients with epilepsy may suffer from severe consequences from seizures (e.g. injuries) when not monitored. Automatic seizure detection systems could mitigate this problem, improving seizure tracking and alerting a caregiver during a seizure. Existing unimodal solutions for seizure detection, based on electroencephalogram (EEG) and electrocardiogram (ECG) still have an unacceptable level of false positives, which can be reduced by combining these two biosignals. In this paper, EEG and ECG data from 7 epileptic patients with diverse recording length and seizure types were used for analyzing the importance of multimodal seizure detection, at a total of around 110 h 2 m. A leave one seizure out cross validation was selected, grouping data containing the period before a seizure and the seizure period. A proof of concept of multimodal seizure detection which uses a deep learning architecture directly on raw data is performed - a Fully Convolutional Neural Network and an architecture based on LSTM were tested. The network based on LSTM achieved better performance - using the best of one or a combination of both signals, all patients had above 91% detected seizures, a specificity per epoch above 0.96 +/- 0.06 and a detection delay below 8.5 +/- 12 s. These results show potential for developing a patient-specific approach for seizure detection that can be transferred to the ambulatory.

Abstract
BACKGROUND: The Medtronic Registry for Epilepsy (MORE; Medtronic Inc, Dublin, Ireland) is an open label observational study evaluating the long-term effectiveness, safety, and performance of deep brain stimulation (DBS) of the anterior nucleus of thalamus (ANT) for the treatment of refractory epilepsy. OBJECTIVE: To compare the difference in success rate of placing contacts at ANT-target region (ANT-TR) between transventricular (TV) and extraventricular (EV) lead trajectories in 73 ANT-DBS implants in 17 European centers participating in the MORE registry. METHODS: The success rate of placing contacts at ANT-TR was evaluated using a screening method combining both individual patient imaging information and stereotactic atlas information to identify contacts at ANT-TR. RESULTS: EV lead trajectory was used in 53% of the trajectories. Approximately, 90% of the TV lead trajectories had at least 1 contact at ANT-TR, vs only 71% of the EV lead trajectories. The success rate for placing at least 1 contact at ANT-TR bilaterally was 84% for TV implants and 58% for EV implants (P <.05; Fisher's exact). No intracranial bleedings were observed, but 1 cortical infarct was reported following EV lead trajectory. CONCLUSION: The results of this registry support the use of TV lead trajectories for ANT-DBS as they have a higher probability in placing contacts at ANT-TR, without appearing to compromise procedural safety. Follow-up data collection is continuing in the MORE registry. These data will provide outcomes associated with TV and EV trajectories.