Automated artifact removal as preprocessing refines neonatal seizure detection.

OBJECTIVE: The description and evaluation of algorithms using Independent Component Analysis (ICA) for automatic removal of ECG, pulsation and respiration artifacts in neonatal EEG before automated seizure detection. METHODS: The developed algorithms decompose the EEG using ICA into its underlying sources. The artifact source was identified using the simultaneously recorded polygraphy signals after preprocessing. The EEG was reconstructed without the corrupting source, leading to a clean EEG. The impact of the artifact removal was measured by comparing the performance of a previously developed seizure detector before and after the artifact removal in 13 selected patients (9 having artifact-contaminated and 4 having artifact-free EEGs). RESULTS: A significant decrease in false alarms (p=0.01) was found while the Good Detection Rate (GDR) for seizures was not altered (p=0.50). CONCLUSIONS: The techniques reduced the number of false positive detections without lowering sensitivity and are beneficial in long term EEG seizure monitoring in the presence of disturbing biological artifacts. SIGNIFICANCE: The proposed algorithms improve neonatal seizure monitoring.

Validation of a new automated neonatal seizure detection system: a clinician's perspective.

OBJECTIVE: To validate an improved automated electroencephalography (EEG)-based neonatal seizure detection algorithm (NeoGuard) in an independent data set. METHODS: EEG background was classified into eight grades based on the evolution of discontinuity and presence of sleep-wake cycles. Patients were further sub-classified into two groups; gpI: mild to moderate (grades 1-5) and gpII: severe (grades 6-8) EEG background abnormalities. Seizures were categorised as definite and dubious. Seizure characteristics were compared between gpI and gpII. The algorithm was tested on 756 h of EEG data from 24 consecutive neonates (median 25 h per patient) with encephalopathy and recorded seizures during continuous monitoring (cEEG). No selection was made regarding the quality of EEG or presence of artefacts. RESULTS: Seizure amplitudes significantly decreased with worsening EEG background. Seizures were detected with a total sensitivity of 61.9% (1285/2077). The detected seizure burden was 66,244/97,574 s (67.9%). Sensitivity per patient was 65.9%, with a mean positive predictive value (PPV) of 73.7%. After excluding four patients with severely abnormal EEG background, and predominantly having dubious seizures, the algorithm showed a median sensitivity per patient of 86.9%, PPV of 89.5% and false positive rate of 0.28 h(-1). Sensitivity tended to be better for patients in gpI. CONCLUSIONS: The algorithm detects neonatal seizures well, has a good PPV and is suited for cEEG monitoring. Changes in electrographic characteristics such as amplitude, duration and rhythmicity in relation to deteriorating EEG background tend to worsen the performance of automated seizure detection. SIGNIFICANCE: cEEG monitoring is important for detecting seizures in the neonatal intensive care unit (NICU). Our automated algorithm reliably detects neonatal seizures that are likely to be clinically most relevant, as reflected by the associated EEG background abnormality.

Time varying neonatal seizure localization.

BACKGROUND: A common cause for damage to the neonatal brain is a shortage in the oxygen supply to the brain or asphyxia. Neonatal seizures are the most frequent manifestation of neonatal neurologic disorders. Multichannel EEG recordings allow topographic localization of seizure foci. OBJECTIVES: We want to objectively determine the spatial distribution of the seizure on the scalp, the location in time and order the dominant sources in the brain based on their strength. METHODS: In this paper we combine a method based on higher order CP-decomposition with subsequent singular value decomposition (SVD). RESULTS: We illustrate the abilities of the method on simulated as well as on real neonatal seizure EEG. CONCLUSIONS: The proposed method provides reliable time and spatial information about the seizure, gives a clear overview of what is going on in the EEG and allows easy interpretation.

Detection algorithm for single motor unit firing in surface EMG of the trapezius muscle.

BACKGROUND: Work-related musculoskeletal disorders (MSD) of the neck and the shoulders are a growing problem in society. An interesting pattern of spontaneous muscle activity, the firing of a single motor unit, in the trapezius muscle is observed during a laboratory study in a rest state or a state with a mental load. OBJECTIVE: In this study, we report on the finding of the single motor unit firing and we present a detection algorithm to localize these single motor unit firings. METHODS: A spike train detection algorithm, using a nonlinear energy operator and correlation, is presented to detect burst of highly correlated, high energetic spike-like segments. RESULTS: This single motor unit was visible in 65% of the test subjects on one or both trapezius muscles although there was no change in posture of the test subjects. All the segments in the data that were determined as single motor unit firings were detected by the algorithm. DISCUSSION: The physiological meaning of this firing pattern is a very low and subconscious contraction of the muscle. A long-term contraction could lead to the exhaustion of the muscle fibers, thus resulting in musculoskeletal disorders. The detection algorithm is able to localize this phenomenon in a sEMG measurement. The ability of detecting these firings is helpful in the research of its origin. CONCLUSION: The detection algorithm can be used to gain insight in the physiological origin of this phenomenon. In addition, the algorithm can also be used in a biofeedback system to warn the user for this undesired contraction to prevent MSD.

Neonatal seizure localization using PARAFAC decomposition.

OBJECTIVE: The description and evaluation of two EEG-based algorithms for automatic and objective determination of the seizure location in the neonatal brain as it is reflected on the scalp. METHODS: Each algorithm extracts the electrical potential distribution of the seizure over the scalp using the higher-order canonical decomposition or Parallel Factor Analysis (PARAFAC), also referred to as the CP model. This model decomposes a tensor in a sum of rank-1 components. The two algorithms differ in the way the tensor is constructed and in the type of activity they are able to extract. While the first method extracts oscillatory seizure activity, the second extracts spike train activity. RESULTS: We compared the seizure localization results of 21 seizures from 6 neonates with post-asphyxial hypoxic ischemic encephalopathy, with that based on the visual analysis of the EEG by a clinical neurophysiologist. There was a good agreement between the two methods in the localization of seizure onset in all. CONCLUSION: The techniques presented in this paper are robust, objective methods to determine neonatal seizure localization. They can be a useful tool for neonatal EEG analysis and for continuous brain function monitoring. SIGNIFICANCE: The proposed algorithms significantly improve neonatal seizure localization and monitoring.

Automated neonatal seizure detection mimicking a human observer reading EEG.

OBJECTIVE: The description and evaluation of a novel patient-independent seizure detection for the EEG of the newborn term infant. METHODS: We identified characteristics of neonatal seizures by which a human observer is able to detect them. Neonatal seizures were divided into two types. For each type, a fully automated detection algorithm was developed based on the identified human observer characteristics. The first algorithm analyzes the correlation between high-energetic segments of the EEG. The second detects increases in low-frequency activity (<8 Hz) with high autocorrelation. RESULTS: The complete algorithm was tested on multi-channel EEG recordings of 21 patients with and 5 patients without electrographic seizures, totaling 217 h of EEG. Sensitivity of the combined algorithms was found to be 88%, Positive Predictive Value (PPV) 75% and the false positive rate 0.66 per hour. CONCLUSIONS: Our approach to separate neonatal seizures into two types yields a high sensitivity combined with a good PPV and much lower false positive rate than previously published algorithms. SIGNIFICANCE: The proposed algorithm significantly improves neonatal seizure detection and monitoring.