Optical Flow Estimation Improves Automated Seizure Detection in Neonatal EEG.

PURPOSE: Existing automated seizure detection algorithms report sensitivities between 43% and 77% and specificities between 56% and 90%. The algorithms suffer from false alarms when applied to neonatal EEG because of the high degree of nurse handling and rhythmic patting used to soothe neonates. Computer vision technology that quantifies movement in real time could distinguish artifactual motion and improve automated neonatal seizure detection algorithms. METHODS: The authors used video EEG recordings from 43 neonates undergoing monitoring for seizures as part of the NEOLEV2 clinical trial. The Persyst neonatal automated seizure detection algorithm ran in real time during study EEG acquisitions. Computer vision algorithms were applied to extract detailed accounts of artifactual movement of the neonate or people near the neonate though dense optical flow estimation. RESULTS: Using the methods mentioned above, 197 periods of patting activity were identified and quantified, of which 45 generated false-positive automated seizure detection events. A binary patting detection algorithm was trained with a subset of 470 event videos. This supervised detection algorithm was applied to a testing subset of 187 event videos with 8 false-positive events, which resulted in a 24% reduction in false-positive automated seizure detections and a 50% reduction in false-positive events caused by neonatal care patting, while maintaining 11 of 12 true-positive seizure detection events. CONCLUSIONS: This work presents a novel approach to improving automated seizure detection algorithms used during neonatal video EEG monitoring. This artifact detection mechanism can improve the ability of a seizure detector algorithm to distinguish between artifact and true seizure activity.

Seizure Detection: Interreader Agreement and Detection Algorithm Assessments Using a Large Dataset.

PURPOSE: To compare the seizure detection performance of three expert humans and two computer algorithms in a large set of epilepsy monitoring unit EEG recordings. METHODS: One hundred twenty prolonged EEGs, 100 containing clinically reported EEG-evident seizures, were evaluated. Seizures were marked by the experts and algorithms. Pairwise sensitivity and false-positive rates were calculated for each human-human and algorithm-human pair. Differences in human pairwise performance were calculated and compared with the range of algorithm versus human performance differences as a type of statistical modified Turing test. RESULTS: A total of 411 individual seizure events were marked by the experts in 2,805 hours of EEG. Mean, pairwise human sensitivities and false-positive rates were 84.9%, 73.7%, and 72.5%, and 1.0, 0.4, and 1.0/day, respectively. Only the Persyst 14 algorithm was comparable with humans-78.2% and 1.0/day. Evaluation of pairwise differences in sensitivity and false-positive rate demonstrated that Persyst 14 met statistical noninferiority criteria compared with the expert humans. CONCLUSIONS: Evaluating typical prolonged EEG recordings, human experts had a modest level of agreement in seizure marking and low false-positive rates. The Persyst 14 algorithm was statistically noninferior to the humans. For the first time, a seizure detection algorithm and human experts performed similarly.

Semiautomated Spike Detection Software Persyst 13 Is Noninferior to Human Readers When Calculating the Spike-Wave Index in Electrical Status Epilepticus in Sleep.

PURPOSE: Our objective was to use semiautomatic methods for calculating the spike-wave index (SWI) in electrical status epilepticus in slow-wave sleep (ESES) and to determine whether this calculation is noninferior to human experts (HEs). METHODS: Each HE marked identical 300-second epochs for all spikes and calculated the SWI in sleep EEGs of patients diagnosed with ESES. Persyst 13 was used to mark spikes (high sensitivity setting) in the same 300-second epochs marked by HEs. The spike-wave index was calculated. Pairwise HE differences and pairwise Persyst 13 (P13)-HE differences for the SWI were calculated. Bootstrap resampling (BCa, N = 3,000) was performed to better estimate mean differences and their 95% confidence bounds between HE and P13-HE pairs. Potential noninferiority of P13 to HEs was tested by comparing the 95% confidence bounds of the mean differences between pairs for the SWI. RESULTS: Twenty EEG records were analyzed. Each HE marked 100 minutes of EEG. HEs 1, 2, 3, and 4 marked 10,075, 8,635, 9,710, and 9,898 spikes, respectively. The highest and lowest 95% confidence bound of the mean difference in the SWI between HE pairs was: High: 10.3%; Low: -10.2%. Highest and lowest 95% confidence bound of the mean difference in the SWI between P13 and HE pairings was as follows: high, 9.5% and low, -6.7%. The lack of a difference between P13 and HEs supports that the algorithm is not inferior to HEs. CONCLUSIONS: Persyst 13 is noninferior to HEs in calculating the SWI in ESES, thus suggesting that an automated approach to SWI calculation may be a useful clinical tool.

Spike detection: Inter-reader agreement and a statistical Turing test on a large data set.

OBJECTIVE: Compare the spike detection performance of three skilled humans and three computer algorithms. METHODS: 40 prolonged EEGs, 35 containing reported spikes, were evaluated. Spikes and sharp waves were marked by the humans and algorithms. Pairwise sensitivity and false positive rates were calculated for each human-human and algorithm-human pair. Differences in human pairwise performance were calculated and compared to the range of algorithm versus human performance differences as a type of statistical Turing test. RESULTS: 5474 individual spike events were marked by the humans. Mean, pairwise human sensitivities and false positive rates were 40.0%, 42.1%, and 51.5%, and 0.80, 0.97, and 1.99/min. Only the Persyst 13 (P13) algorithm was comparable to humans - 43.9% and 1.65/min. Evaluation of pairwise differences in sensitivity and false positive rate demonstrated that P13 met statistical noninferiority criteria compared to the humans. CONCLUSION: Humans had only a fair level of agreement in spike marking. The P13 algorithm was statistically noninferior to the humans. SIGNIFICANCE: This was the first time that a spike detection algorithm and humans performed similarly. The performance comparison methodology utilized here is generally applicable to problems in which skilled human performance is the desired standard and no external gold standard exists.

Algorithm architectures for patient dependent seizure detection.

OBJECTIVE: The goal of this work is to determine whether improved performance (compared to patient independent algorithms) can be achieved by an algorithm, developed on the fly, that requires no user input beyond the identification of the first one or two seizures in the record. METHODS: The previously developed AutoLearn algorithm, which employs the probabilistic neural network (PNN), is tested on 209 seizures obtained from the epilepsy monitoring unit (EMU) or ambulatory recordings. A construction algorithm is used to compare a variety of algorithm architectures and factors. The Taguchi design of experiments (DoE) method is employed find the significant factors without resorting to a full factorial design. RESULTS: Architectures that train a single PNN per channel and use segmentation to identify ranges of similar activity are preferred. The two best architectures are insensitive to the levels of any of the other factors tested. The training time for the algorithm is less than 1s, and approximately 2 min are required to find the seizures in an 8 h record. CONCLUSIONS: The final algorithm, which requires no input from the user other than the marking of the first seizure in a record, performs as well or better than the 3 seizure detectors on EMU and ambulatory records. The algorithm performs nearly as well as human experts on the EMU records. SIGNIFICANCE: The described method can be used to identify unusual seizures (or other patterns) that will be missed by the current generation of seizure detectors. We expect that the methods developed here will also aid the development of patient independent seizure detectors that can improve their performance over time by incorporating new examples.

A neural network method for automatic and incremental learning applied to patient-dependent seizure detection.

OBJECTIVE: Describe and evaluate a neural network method for automatic and incremental learning applied to patient-dependent seizure detection. Compare the classification ability of various time-frequency methods including FFT spectrogram, spectral edge frequency and bicoherence. METHODS: 57 seizures from 10 epilepsy patients are used. A probabilistic neural network (PNN) is trained and incrementally updated in a novel fashion. The speed and accuracy of the method is evaluated with different training parameters and time-frequency methods. RESULTS: Training the PNN on a single seizure from each record offers better performance (sensitivity = 0.89 and false-positive-rate = 0.56/h) than 3 patient-independent seizure detection algorithms. The method is virtually unaffected by the settings of various training parameters. Training is very fast (0.9 s), and the accuracy improves as more examples are added incrementally (without retraining). The overall best time-frequency method was the FFT spectrogram. The bicoherence plus the FFT spectrogram was the best method on 4 records, improving the correlation from 0.111 to 0.940 on one and from 0.288 to 0.612 on another. CONCLUSIONS: The proposed method offers accurate, robust and virtually instantaneous training and incremental learning when applied to patient-dependent seizure detection. SIGNIFICANCE: Accurate seizure detection can improve patient care in both the epilepsy monitoring unit and the intensive care unit. Future applications include patient-independent algorithms that continue to learn as new examples are encountered.

Data analysis for continuous EEG monitoring in the ICU: seeing the forest and the trees.

Continuous EEG monitoring (CEEG) is a powerful tool for evaluating cerebral function in obtunded and comatose critically ill patients. The ongoing analysis of CEEG data is a major task because of the volume of data generated during monitoring and the need for near real-time interpretation of a patient's EEG patterns. Advances in digital EEG data acquisition, computer processing, data transmission, and data display have made CEEG monitoring in the intensive care unit technically feasible. A variety of quantitative EEG tools such as Fourier analysis and amplitude-integrated EEG, and other methods of data analysis such as computerized seizure detection, increasingly allow for focused review of EEG epochs of potential interest. These tools reduce the tremendous time burdens that accompany analysis of the complete CEEG data stream, and allow bedside personnel and nonexpert staff to potentially recognize significant EEG changes in a timely fashion. This article uses literature review and clinical case examples to illustrate techniques for the display and analysis of intensive care unit CEEG recordings. Areas requiring further research and development are discussed.

Seizure detection: evaluation of the Reveal algorithm.

OBJECTIVE: The aim of this study is to evaluate an improved seizure detection algorithm and to compare with two other algorithms and human experts. METHODS: 672 seizures from 426 epilepsy patients were examined with the (new) Reveal algorithm which utilizes 3 methods, novel in their application to seizure detection: Matching Pursuit, small neural network-rules and a new connected-object hierarchical clustering algorithm. RESULTS: Reveal had a sensitivity of 76% with a false positive rate of 0.11/h. Two other algorithms (Sensa and CNet) were tested and had sensitivities of 35.4 and 48.2% and false positive rates of 0.11/h and 0.75/h, respectively. CONCLUSIONS: This study validates the Reveal algorithm, and shows it to compare favorably with other methods. SIGNIFICANCE: Improved seizure detection can improve patient care in both the epilepsy monitoring unit and the intensive care unit.

Seizure detection: correlation of human experts.

OBJECTIVE: The description and application of a new, overlap-integral comparison method and the quantification of human vs. human accuracies that can be used as goals for algorithms. METHODS: Four human experts marked ten 8 h electroencephalography (EEG) records from seizure patients. The seizures varied in origin and type, including complex partial, generalized absence, secondarily generalized and primary generalized tonic-clonic. The traditional any-overlap comparison method is used in addition to the overlap-integral method, which is sensitive to the correct placement of the seizure endpoints. RESULTS: The number of events marked by each reader ranged from 57 to 77. The average any-overlap sensitivity and false positives per hour rate are 0.92 and 0.117. The average overlap-integral correlation, sensitivity and specificity are 0.80, 0.82 and 0.9926. As expected, the correspondence between readers is high, but confounding issues resulted in overlap-integral sensitivities less than 0.5 for 10% of the records. Seven percent of the any-overlap sensitivities are less than 0.5. A comparison of the methods by record shows that the overlap-integral specificity and the any-overlap false positive rate measure different features. CONCLUSIONS: There was little variation between readers and they were essentially interchangeable. High seizure rate (many per hour), short seizure durations (<10 s) and long seizure durations (approximately 10 min) with ambiguous offsets can complicate the analysis and result in poor correlation. There may be any number of unmarked events in rigorously marked records and it may be preferable to use records from non-epilepsy patients to compute the false positive rate. The any-overlap and overlap-integral comparison methods are complementary. SIGNIFICANCE: Correlation between expert human readers can be low on some records, which will complicate testing of seizure detection algorithms.

Spike detection: a review and comparison of algorithms.

For algorithm developers, this review details recent approaches to the problem, compares the accuracy of various algorithms, identifies common testing issues and proposes some solutions. For the algorithm user, e.g. electroencephalograph (EEG) technician or neurologist, this review provides an estimate of algorithm accuracy and comparison to that of human experts. Manuscripts dated from 1975 are reviewed. Progress since Frost's 1985 review of the state of the art is discussed. Twenty-five manuscripts are reviewed. Many novel methods have been proposed including neural networks and high-resolution frequency methods. Algorithm accuracy is less than that of experts, but the accuracy of experts is probably less than what is commonly believed. Larger record sets will be required for expert-level detection algorithms.

Introduction to hierarchical clustering.

Hierarchical clustering of spike events is a method of grouping events that are similar in topology, morphology, or both, and it provides a method of efficient, detailed analysis of interictal events. Information about the relative populations of spikes at multiple foci is presented, and artifact events are grouped and eliminated en masse. The process of hierarchical clustering is explained, and a set of simulated traces is used to illustrate the process of hierarchical clustering and the development of a cluster tree to display the relative populations of similar spike events. Using EEG data from long-term monitoring, the use of a "review wizard" is explored as a means of structuring the process of hierarchical clustering and traversing the cluster tree. This aid is also used to streamline the process of determining the similarity of events within each group and of verifying that events exhibiting clinically important differences are not hidden within the groups comprising the average traces.