Unsupervised Classification of High-Frequency Oscillations in Human Neocortical Epilepsy and Control Patients

High-frequency oscillations (HFOs) have been observed in animal and human intracranial recordings during both normal and aberrant brain states. It has been proposed that the relationship between subclasses of these oscillations can be used to identify epileptic brain. Studies of HFOs in epilepsy have been hampered by selection bias arising primarily out of the need to reduce the volume of data so that clinicians can manually review it. In this study, we introduce an algorithm for detecting and classifying these signals automatically and demonstrate the tractability of analyzing a data set of unprecedented size, over 31,000 channel-hours of intracranial electroencephalographic (iEEG) recordings from micro- and macroelectrodes in humans. Using an unsupervised approach that does not presuppose a specific number of clusters in the data, we show direct evidence for the existence of distinct classes of transient oscillations within the 100- to 500-Hz frequency range in a population of nine neocortical epilepsy patients and two controls. The number of classes we find, four (three plus one putative artifact class), is consistent with prior studies that identify “ripple” and “fast ripple” oscillations using human-intensive methods and, additionally, identifies a less examined class of mixed-frequency events.

Download Full-text

Towards automatic classification of pathological epileptic tissue with interictal high frequency oscillations

10.1101/2020.05.28.122416 ◽

2020 ◽

Author(s):

Michael D. Nunez ◽

Krit Charupanit ◽

Indranil Sen-Gupta ◽

Beth A. Lopour ◽

Jack J. Lin

Keyword(s):

High Frequency ◽

Negative Binomial ◽

Slow Wave Sleep ◽

Nrem Sleep ◽

Oscillatory Dynamics ◽

Scalp Eeg ◽

High Frequency Oscillations ◽

Binomial Mixture ◽

Brain States

AbstractHigh frequency oscillations (HFOs) recorded by intracranial electrodes have generated excitement for their potential to help localize epileptic tissue for surgical resection (Frauscher et al., 2017). However, previous research has shown that the number of HFOs per minute (i.e. the HFO “rate”) is not stable over the duration of intracranial recordings. The rate of HFOs increases during periods of slow-wave sleep (von Ellenrieder et al., 2017), and HFOs that are predictive of epileptic tissue may occur in oscillatory patterns (Motoi et al., 2018). We sought to further understand how between-seizure (i.e. “interictal”) HFO dynamics predict the seizure onset zone (SOZ). Using long-term intracranial EEG from 16 subjects, we fit Poisson and Negative Binomial mixture models that describe HFO dynamics and include the ability to switch between two discrete brain states. Oscillatory dynamics of HFO occurrences were found to be predictive of SOZ and were more consistently predictive than HFO rate. Using concurrent scalp-EEG in two patients, we show that the model-found brain states corresponded to (1) non-REM (NREM) sleep and (2) awake and rapid eye movement (REM) sleep. This work suggests that unsupervised approaches for classification of epileptic tissue without sleep-staging can be developed using mixture modeling of HFO dynamics.

Download Full-text

Patient-specific optimization of automated detection improves seizure onset zone localization based on high frequency oscillations

10.1101/2020.09.14.297309 ◽

2020 ◽

Author(s):

Casey L. Trevino ◽

Jack J. Lin ◽

Indranil Sen-Gupta ◽

Beth A. Lopour

Keyword(s):

High Frequency ◽

Classification Accuracy ◽

Patient Specific ◽

Seizure Onset ◽

Data Set ◽

Localization Accuracy ◽

High Frequency Oscillations ◽

Seizure Onset Zone ◽

Wide Range ◽

The Impact

AbstractHigh frequency oscillations (HFOs) are a promising biomarker of epileptogenicity, and automated algorithms are critical tools for their detection. However, previously validated algorithms often exhibit decreased HFO detection accuracy when applied to a new data set, if the parameters are not optimized. This likely contributes to decreased seizure localization accuracy, but this has never been tested. Therefore, we evaluated the impact of parameter selection on seizure onset zone (SOZ) localization using automatically detected HFOs. We detected HFOs in intracranial EEG from twenty medically refractory epilepsy patients with seizure free surgical outcomes using an automated algorithm. For each patient, we assessed classification accuracy of channels inside/outside the SOZ using a wide range of detection parameters and identified the parameters associated with maximum classification accuracy. We found that only three out of twenty patients achieved maximal localization accuracy using conventional HFO detection parameters, and optimal parameter ranges varied significantly across patients. The parameters for amplitude threshold and root-mean-square window had the greatest impact on SOZ localization accuracy; minimum event duration and rejection of false positive events did not significantly affect the results. Using individualized optimal parameters led to substantial improvements in localization accuracy, particularly in reducing false positives from non-SOZ channels. We conclude that optimal HFO detection parameters are patient-specific, often differ from conventional parameters, and have a significant impact on SOZ localization. This suggests that individual variability should be considered when implementing automatic HFO detection as a tool for surgical planning.

Download Full-text

Text detection in images based on unsupervised classification of high-frequency wavelet coefficients

Proceedings of the 17th International Conference on Pattern Recognition, 2004. ICPR 2004. ◽

10.1109/icpr.2004.1334146 ◽

2004 ◽

Cited By ~ 95

Author(s):

J. Gllavata ◽

R. Ewerth ◽

B. Freisleben

Keyword(s):

High Frequency ◽

Unsupervised Classification ◽

Text Detection ◽

Wavelet Coefficients

Download Full-text

Postoperative seizure outcome-guided machine learning for interictal electrocorticography in neocortical epilepsy

Journal of Neurophysiology ◽

10.1152/jn.00225.2017 ◽

2018 ◽

Vol 119 (6) ◽

pp. 2265-2275 ◽

Cited By ~ 2

Author(s):

Seong-Cheol Park ◽

Chun Kee Chung

Keyword(s):

Machine Learning ◽

Free Group ◽

High Frequency ◽

Low Frequency ◽

Outcome Data ◽

Training Data ◽

Seizure Outcome ◽

Data Set ◽

Neocortical Epilepsy ◽

Postoperative Seizure

The objective of this study was to introduce a new machine learning guided by outcome of resective epilepsy surgery defined as the presence/absence of seizures to improve data mining for interictal pathological activities in neocortical epilepsy. Electrocorticographies for 39 patients with medically intractable neocortical epilepsy were analyzed. We separately analyzed 38 frequencies from 0.9 to 800 Hz including both high-frequency activities and low-frequency activities to select bands related to seizure outcome. An automatic detector using amplitude-duration-number thresholds was used. Interictal electrocorticography data sets of 8 min for each patient were selected. In the first training data set of 20 patients, the automatic detector was optimized to best differentiate the seizure-free group from not-seizure-free-group based on ranks of resection percentages of activities detected using a genetic algorithm. The optimization was validated in a different data set of 19 patients. There were 16 (41%) seizure-free patients. The mean follow-up duration was 21 ± 11 mo (range, 13–44 mo). After validation, frequencies significantly related to seizure outcome were 5.8, 8.4–25, 30, 36, 52, and 75 among low-frequency activities and 108 and 800 Hz among high-frequency activities. Resection for 5.8, 8.4–25, 108, and 800 Hz activities consistently improved seizure outcome. Resection effects of 17–36, 52, and 75 Hz activities on seizure outcome were variable according to thresholds. We developed and validated an automated detector for monitoring interictal pathological and inhibitory/physiological activities in neocortical epilepsy using a data-driven approach through outcome-guided machine learning. NEW & NOTEWORTHY Outcome-guided machine learning based on seizure outcome was used to improve detections for interictal electrocorticographic low- and high-frequency activities. This method resulted in better separation of seizure outcome groups than others reported in the literature. The automatic detector can be trained without human intervention and no prior information. It is based only on objective seizure outcome data without relying on an expert’s manual annotations. Using the method, we could find and characterize pathological and inhibitory activities.

Download Full-text

Classification of Ancient Roman Coins by Denomination Using Colour, a Forgotten Feature in Automatic Ancient Coin Analysis

Sci ◽

10.3390/sci2020037 ◽

2020 ◽

Vol 2 (2) ◽

pp. 37

Author(s):

Yuanyuan Ma ◽

Ognjen Arandjelović

Keyword(s):

Relevant Information ◽

Data Set ◽

Roman Coins ◽

Methodological Approaches ◽

Ancient Coin ◽

Number Of Classes ◽

First Time ◽

Enormous Number ◽

Potential Use

Ancient numismatics, that is, the study of ancient currencies (predominantly coins), is an interesting domain for the application of computer vision and machine learning, and has been receiving an increasing amount of attention in recent years. Notwithstanding the number of articles published on the topic, the variety of different methodological approaches described, and the mounting realisation that the relevant problems in the field are most challenging indeed, all research to date has entirely ignored one specific, readily accessible modality: colour. Invariably, colour is discarded and images of coins treated as being greyscale. The present article is the first one to question this decision (and indeed, it is a decision). We discuss the reasons behind the said choice, present a case why it ought to be reexamined, and in turn investigate the issue for the first time in the published literature. Specifically, we propose two new colour-based representations specifically designed with the aim of being applied to ancient coin analysis, and argue why it is sensible to employ them in the first stages of the classification process as a means of drastically reducing the initially enormous number of classes involved in type matching ancient coins (tens of thousands, just for Ancient Roman Imperial coins). Furthermore, we introduce a new data set collected with the specific aim of denomination-based categorisation of ancient coins, where we hypothesised colour could be of potential use, and evaluate the proposed representations. Lastly, we report surprisingly successful performances which goes further than confirming our hypothesis—rather, they convincingly demonstrate a much higher relevant information content carried by colour than even we expected. Thus we trust that our findings will be noted by others in the field and that more attention and further research will be devoted to the use of colour in automatic ancient coin analysis.

Download Full-text