scholarly journals Recognition of Epileptic Seizures in EEG Records: A Transfer Learning Approach

2021 ◽  
Vol 10 (1) ◽  
pp. 61
Author(s):  
Elena Caires Silveira ◽  
Caio Fellipe Santos Corrêa

Introduction: Seizure is a transient phenomenon with genesis in excessive abnormal or synchronous neuronal electrical activity in the brain, while epilepsy is defined as a brain dysfunction characterized by persistent predisposition to generate seizures. The identification of epileptogenic electroencephalographic patterns can be performed using machine learning.the present study aimed to develop a transfer learning based classifier able to detect epileptic seizures in images generated from electroencephalographic data graphic representation.Material and Methods: We used the Epileptic Seizure Recognition Data Set,which consists of 500 brain activity records for 23.6 seconds comprising 23 chunks of 178 data points, and transformed the resulting 11500 instances into images by graphically plotting its data points. Those images were then splitted in training and test set and used to build and assess, respectvely, a transfer learning-based deep neural network, which classified the images according the presence or absence of epileptic seizures.Results: The model achieved 100% accuracy, sensitivity and specificity, with a AUC-score of 1.0, demonstrating the great potential of transfer learning for the analysis of graphically represented electroencephalographic data.Conclusion: It is opportune to raise new studies involving transfer learning for the analysis of signal data, with the aim of improving, disseminating and validating its use for daily clinical practice.

2010 ◽  
Vol 24 (2) ◽  
pp. 131-135 ◽  
Author(s):  
Włodzimierz Klonowski ◽  
Pawel Stepien ◽  
Robert Stepien

Over 20 years ago, Watt and Hameroff (1987 ) suggested that consciousness may be described as a manifestation of deterministic chaos in the brain/mind. To analyze EEG-signal complexity, we used Higuchi’s fractal dimension in time domain and symbolic analysis methods. Our results of analysis of EEG-signals under anesthesia, during physiological sleep, and during epileptic seizures lead to a conclusion similar to that of Watt and Hameroff: Brain activity, measured by complexity of the EEG-signal, diminishes (becomes less chaotic) when consciousness is being “switched off”. So, consciousness may be described as a manifestation of deterministic chaos in the brain/mind.


Author(s):  
V. A. Maksimenko ◽  
A. A. Harchenko ◽  
A. Lüttjohann

Introduction: Now the great interest in studying the brain activity based on detection of oscillatory patterns on the recorded data of electrical neuronal activity (electroencephalograms) is associated with the possibility of developing brain-computer interfaces. Braincomputer interfaces are based on the real-time detection of characteristic patterns on electroencephalograms and their transformation  into commands for controlling external devices. One of the important areas of the brain-computer interfaces application is the control of the pathological activity of the brain. This is in demand for epilepsy patients, who do not respond to drug treatment.Purpose: A technique for detecting the characteristic patterns of neural activity preceding the occurrence of epileptic seizures.Results:Using multi-channel electroencephalograms, we consider the dynamics of thalamo-cortical brain network, preceded the occurrence of an epileptic seizure. We have developed technique which allows to predict the occurrence of an epileptic seizure. The technique has been implemented in a brain-computer interface, which has been tested in-vivo on the animal model of absence epilepsy.Practical relevance:The results of our study demonstrate the possibility of epileptic seizures prediction based on multichannel electroencephalograms. The obtained results can be used in the development of neurointerfaces for the prediction and prevention of seizures of various types of epilepsy in humans. 


2018 ◽  
Vol 29 (8) ◽  
pp. 825-835 ◽  
Author(s):  
Sergei V. Fedorovich ◽  
Tatyana V. Waseem

AbstractBrain tissue is bioenergetically expensive. In humans, it composes approximately 2% of body weight and accounts for approximately 20% of calorie consumption. The brain consumes energy mostly for ion and neurotransmitter transport, a process that occurs primarily in synapses. Therefore, synapses are expensive for any living creature who has brain. In many brain diseases, synapses are damaged earlier than neurons start dying. Synapses may be considered as vulnerable sites on a neuron. Ischemic stroke, an acute disturbance of blood flow in the brain, is an example of a metabolic disease that affects synapses. The associated excessive glutamate release, called excitotoxicity, is involved in neuronal death in brain ischemia. Another example of a metabolic disease is hypoglycemia, a complication of diabetes mellitus, which leads to neuronal death and brain dysfunction. However, synapse function can be corrected with “bioenergetic medicine”. In this review, a ketogenic diet is discussed as a curative option. In support of a ketogenic diet, whereby carbohydrates are replaced for fats in daily meals, epileptic seizures can be terminated. In this review, we discuss possible metabolic sensors in synapses. These may include molecules that perceive changes in composition of extracellular space, for instance, ketone body and lactate receptors, or molecules reacting to changes in cytosol, for instance, KATPchannels or AMP kinase. Inhibition of endocytosis is believed to be a universal synaptic mechanism of adaptation to metabolic changes.


2019 ◽  
Vol 18 ◽  
pp. 153303381985836 ◽  
Author(s):  
Quan Chen ◽  
Shiliang Hu ◽  
Peiran Long ◽  
Fang Lu ◽  
Yujie Shi ◽  
...  

Purpose: In prostate focal therapy, it is important to accurately localize malignant lesions in order to increase biological effect of the tumor region while achieving a reduction in dose to noncancerous tissue. In this work, we proposed a transfer learning–based deep learning approach, for classification of prostate lesions in multiparametric magnetic resonance imaging images. Methods: Magnetic resonance imaging images were preprocessed to remove bias artifact and normalize the data. Two state-of-the-art deep convolutional neural network models, InceptionV3 and VGG-16, were pretrained on ImageNet data set and retuned on the multiparametric magnetic resonance imaging data set. As lesion appearances differ by the prostate zone that it resides in, separate models were trained. Ensembling was performed on each prostate zone to improve area under the curve. In addition, the predictions from lesions on each prostate zone were scaled separately to increase the area under the curve for all lesions combined. Results: The models were tuned to produce the highest area under the curve on validation data set. When it was applied to the unseen test data set, the transferred InceptionV3 model achieved an area under the curve of 0.81 and the transferred VGG-16 model achieved an area under the curve of 0.83. This was the third best score among the 72 methods from 33 participating groups in ProstateX competition. Conclusion: The transfer learning approach is a promising method for prostate cancer detection on multiparametric magnetic resonance imaging images. Features learned from ImageNet data set can be useful for medical images.


Author(s):  
Rekh Ram Janghel ◽  
Yogesh Kumar Rathore ◽  
Gautam Tatiparti

Epilepsy is a brain ailment identified by unpredictable interruptions of normal brain activity. Around 1% of mankind experience epileptic seizures. Around 10% of the United States population experiences at least a single seizure in their life. Epilepsy is distinguished by the tendency of the brain to generate unexpected bursts of unusual electrical activity that disrupts the normal functioning of the brain. As seizures usually occur rarely and are unforeseeable, seizure recognition systems are recommended for seizure detection during long-term electroencephalography (EEG). In this chapter, ANN models, namely, BPA, RNN, CL, PNN, and LVQ, have been implemented. A prominent dataset was employed to assess the proposed method. The proposed method is capable of achieving an accuracy of 97.5%; the high accuracy obtained has confirmed the great success of the method.


Data in Brief ◽  
2021 ◽  
pp. 106993
Author(s):  
Daisuke Nishida ◽  
Katsuhiro Mizuno ◽  
Emi Yamada ◽  
Tetsuya Tsuji ◽  
Takashi Hanakawa ◽  
...  

Author(s):  
Giuseppe Canonaco ◽  
Manuel Roveri ◽  
Cesare Alippi ◽  
Fabrizio Podenzani ◽  
Antonio Bennardo ◽  
...  

AbstractPipeline infrastructures, carrying either gas or oil, are often affected by internal corrosion, which is a dangerous phenomenon that may cause threats to both the environment (due to potential leakages) and the human beings (due to accidents that may cause explosions in presence of gas leakages). For this reason, predictive mechanisms are needed to detect and address the corrosion phenomenon. Recently, we have seen a first attempt at leveraging Machine Learning (ML) techniques in this field thanks to their high ability in modeling highly complex phenomena. In order to rely on these techniques, we need a set of data, representing factors influencing the corrosion in a given pipeline, together with their related supervised information, measuring the corrosion level along the considered infrastructure profile. Unfortunately, it is not always possible to access supervised information for a given pipeline since measuring the corrosion is a costly and time-consuming operation. In this paper, we will address the problem of devising a ML-based predictive model for internal corrosion under the assumption that supervised information is unavailable for the pipeline of interest, while it is available for some other pipelines that can be leveraged through Transfer Learning (TL) to build the predictive model itself. We will cover all the methodological steps from data set creation to the usage of TL. The whole methodology will be experimentally validated on a set of real-world pipelines.


Webology ◽  
2020 ◽  
Vol 17 (2) ◽  
pp. 717-734
Author(s):  
G. Maragatham ◽  
T. Kirthiga Devi ◽  
P. Savaridassan ◽  
Sachin Garg

Epilepsy is a neurological disorder that disturbs the brain and causes abnormal brain activity. It results in loss of awareness in some cases and unusual behavior and sensations. In this regard, if the seizures could be identified in its earlier stages then the patient can be provided appropriate care and treatment in time and prevent any severe damage to the patient as a whole. In this paper, we try to detect epilepsy using the EEG Signal Recordings and classify them using pre-trained CNN models between preictal and interictal classes. For this we are advocating the use of American Society for Epilepsy Dataset. The focus is on detecting the epilepsy pattern from the EEG recordings in a timely and accurate manner.


2019 ◽  
Vol 6 (4) ◽  
pp. 1773
Author(s):  
Bella Kurnia

Epilepsy is a syndrome of brain dysfunction induced by the aberrant excitability of certain neurons. Despite advances in surgical technique and anti-epileptic drug in recent years, recurrent epileptic seizures remain intractable and lead to a serious morbidity in the world. The ketogenic diet (KD) is a nonpharmacologic treatment that has been used for refractory epilepsy since 1921. The KD is a high-fat, low-carbohydrate, and restricted protein diet, which is calculated and weighed for each individual patient. The goal of the KD treatment is to bring the brain into a state of ketosis to control seizures. Many studies have shown that ketogenic diet was very useful in controlling refractory epilepsy.


Sign in / Sign up

Export Citation Format

Share Document