Detecting epileptic seizure from scalp EEG using Lyapunov spectrum

Using Explainable Artificial Intelligence to Obtain Efficient Seizure-Detection Models Based on Electroencephalography Signals

Epilepsy is a condition that affects 50 million individuals globally, significantly impacting their quality of life. Epileptic seizures, a transient occurrence, are characterized by a spectrum of manifestations, including alterations in motor function and consciousness. These events impose restrictions on the daily lives of those affected, frequently resulting in social isolation and psychological distress. In response, numerous efforts have been directed towards the detection and prevention of epileptic seizures through EEG signal analysis, employing machine learning and deep learning methodologies. This study presents a methodology that reduces the number of features and channels required by simpler classifiers, leveraging Explainable Artificial Intelligence (XAI) for the detection of epileptic seizures. The proposed approach achieves performance metrics exceeding 95% in accuracy, precision, recall, and F1-score by utilizing merely six features and five channels in a temporal domain analysis, with a time window of 1 s. The model demonstrates robust generalization across the patient cohort included in the database, suggesting that feature reduction in simpler models-without resorting to deep learning-is adequate for seizure detection. The research underscores the potential for substantial reductions in the number of attributes and channels, advocating for the training of models with strategically selected electrodes, and thereby supporting the development of effective mobile applications for epileptic seizure detection.

Prediction of Seizure Recurrence. A Note of Caution

Great strides have been made recently in documenting that machine-learning programs can predict seizure occurrence in people who have epilepsy. Along with this progress have come claims that appear to us to be a bit premature. We anticipate that many people will benefit from seizure prediction. We also doubt that all will benefit. Although machine learning is a useful tool for aiding discovery, we believe that the greatest progress will come from deeper understanding of seizures, epilepsy, and the EEG features that enable seizure prediction. In this essay, we lay out reasons for optimism and skepticism.

Epileptic Seizure Detection Using Multi-Channel EEG Wavelet Power Spectra and 1-D Convolutional Neural Networks

The use of feature extraction and selection from EEG signals has shown to be useful in the detection of epileptic seizure segments. However, these traditional methods have more recently been surpassed by deep learning techniques, forgoing the need for complex feature engineering. This work aims to extend the conventional approach of epileptic seizure detection utilizing raw power spectra of EEG signals and convolutional neural networks (CNN). The proposed technique utilizes wavelet transform to compute the frequency characteristics of multi-channel EEG signals. The EEG signals are divided into 2 second epochs and frequency spectrum up to a cutoff frequency of 45 Hz is computed. This multi-channel raw spectral data forms the input to a one-dimensional CNN (1-D CNN). Spectral data from the current, previous, and next epochs is utilized for predicting the label of the current epoch. The performance of the technique is evaluated using a dataset of EEG signals from 24 cases. The proposed method achieves an accuracy of 97.25% in detecting epileptic seizure segments. This result shows that multi-channel EEG wavelet power spectra and 1-D CNN are useful in detecting epileptic seizures.

EEG Analysis in Structural Focal Epilepsy Using the Methods of Nonlinear Dynamics (Lyapunov Exponents, Lempel-Ziv Complexity, and Multiscale Entropy)

This paper analyzes a case with the patient having focal structural epilepsy by processing electroencephalogram (EEG) fragments containing the "sharp wave" pattern of brain activity. EEG signals were recorded using 21 channels. Based on the fact that EEG signals are time series, an approach has been developed for their analysis using nonlinear dynamics tools: calculating the Lyapunov exponent's spectrum, multiscale entropy, and Lempel-Ziv complexity. The calculation of the first Lyapunov exponent is carried out by three methods: Wolf, Rosenstein, and Sano-Sawada, to obtain reliable results. The seven Lyapunov exponent spectra are calculated by the Sano-Sawada method. For the observed patient, studies showed that with medical treatment, his condition did not improve, and as a result, it was recommended to switch from conservative treatment to surgical. The obtained results of the patient's EEG study using the indicated nonlinear dynamics methods are in good agreement with the medical report and MRI data. The approach developed for the analysis of EEG signals by nonlinear dynamics methods can be applied for early detection of structural changes.

Significance of using a nonlinear analysis technique, the Lyapunov exponent, on the understanding of the dynamics of the cardiorespiratory system in rats

BACKGROUND/AIM

Pneumocardiography (PNCG) is the recording method of cardiac-induced tracheal air flow and pressure pulsations in the respiratory airways. PNCG signals reflect both the lung and heart actions and could be accurately recorded in spontaneously breathing anesthetized rats. Nonlinear analysis methods, including the Lyapunov exponent, can be used to explain the biological dynamics of systems such as the cardiorespiratory system.

MATERIALS AND METHODS

In this study, we recorded tracheal air flow signals, including PNCG signals, from 3 representative anesthetized rats and analyzed the nonlinear behavior of these complex signals using Lyapunov exponents.

RESULTS

Lyapunov exponents may also be used to determine the normal and pathological structure of biological systems. If the signals have at least one positive Lyapunov exponent, the signals reflect chaotic activity, as seen in PNCG signals in rats; the largest Lyapunov exponents of the signals of the healthy rats were greater than zero in this study.

CONCLUSION

A method was proposed to determine the diagnostic and prognostic values of the cardiorespiratory system of rats using the arrangement of the PNCG and Lyapunov exponents, which may be monitored as vitality indicators.

A primary malignant fibrous histiocytoma of the scalp and intracranial tumor bleeding: a case report

INTRODUCTION

A malignant fibrous histiocytoma occurring on the scalp near a primary operation site is extremely rare.

CASE PRESENTATION

A 74-year-old Chinese man presented with a one-month history of recurrent headaches, vomiting and left limb atony. He had undergone a successful clipping operation through the pterional approach for an anterior communicating aneurysm five years previously. One month before presentation, he developed a headache without apparent cause that was especially severe in the right frontal region. He also had a small tumor on the right side of his forehead at the original incision site. The tumor had gradually increased from soybean size to egg size in one month; this growth was accompanied by nausea, projectile vomiting of gastric contents and left limb atony. The subcutaneous tumor was totally resected along with some affected cranial tissues. Our patient's postoperative recovery was good, and he was safely discharged 20 days after the surgery. He was free of recurrence over two years of follow-up.

CONCLUSIONS

Dissection of the temporal muscles and deep fascia using electric resection and electrocoagulation through the pterional approach may cause tissue degeneration, which may in turn lead to cancer development. In our patient's case, the reason for the development of the tumor five years after his surgical aneurysm repair was unclear; it may have represented a primary malignant fibrous histiocytoma of the scalp that had no relationship to the operation. We followed up our patient for two years and he had no tumor recurrence. Because malignant fibrous histiocytoma of the scalp has a high degree of malignancy and readily recurs in situ, early diagnosis and radical surgical resection are key to a successful outcome.