Amazing technological breakthrough possible @S-Logix

Office Address

  • #5, First Floor, 4th Street Dr. Subbarayan Nagar Kodambakkam, Chennai-600 024 Landmark : Samiyar Madam
  • +91- 81240 01111

Social List

Research Proposal on Deep Learning Models for Epileptic Focus Localization

Research Proposal on Deep Learning Models for Epileptic Focus Localization

  Epileptic focus localization plays a crucial role in epilepsy diagnosis and manages the surgical interventions in the brain of patients with epilepsy. The electrographic indications of focal epilepsy depend upon various aspects, such as the size and location of the seizure generator, the area and number of recording electrodes, and the diminishing characteristics of the skull and other interceding tissues. In epileptic surgery, epileptic focus localization aims to identify and detach the seizure onset area from the brain.
  With the development of computational methods such as Artificial Intelligence (AI), epileptic foci localization still has an interesting scope for an epilepsy diagnosis. Epileptic focus localization is a binary classification problem differentiating epileptic EEG signals into focal and non-focal EEG signals. Localization of epileptic seizure focus via visual perception by healthcare experts is time-consuming and error-prone.
  Several conventional methods have been developed for epileptic focus localization, including biomarker-based methods, statistical feature extraction, and neural networks. Utilizing the machine learning models for the epileptic focus localization faces issues, including high computation time due to complex feature extraction and expensive generalization to handle huge data.

Current Trends of Deep Learning in Epileptic Focal Localization:
• Advanced machine learning called deep learning models is employed for automatic, efficient, and accurate localization of focal seizures.
• Deep learning models automatically extract the discriminative features from epileptic or intracranial EEG signals to classify them into focal and non-focal.
• The most commonly exploited deep learning models for epileptic focus localization are autoencoders (AE) and convolutional neural networks (CNN).
• CNN-based epileptic foci detection utilizes time-frequency feature fusion to enhance epileptic focus detection rate and lessen the treatment time of epilepsy.
• Multi-scale concept deep learning strategy employs the CNN model to localize the epileptic foci using Positron Emission Tomography (PET) scans.
• Integrating semi-supervised learning and unsupervised learning framework exploited for accurate automatic epileptic focus localization with the help of deep learning models and electroencephalogram signals.
• More recently, hybrid features, machine learning classifiers, and deep neural networks are combined to detect epileptogenic zone localization, which is beneficial for developing smart Internet of Medical Things (IoMT) devices.

Open Opportunities of Epileptic Focal Localization: Even though the deep learning model assisted in epileptic surgery to produce effective epileptic focus localization that helps the surgeons to remove the affected epileptic focal area in the brain. Some research issues that must be solved to impart better computational outcomes for localizing epileptic zone;
• The variability among epileptic patients should be considered in further developing epileptic foci localization using deep learning.
• For real-world healthcare applications, designing a patient-independent system for identifying seizure onset zone is necessary to develop for varying electrodes and the patient-specific nature of EEG signals.
• As the promising future direction of epileptic foci localization adaptation to different distributions involves, transfer learning and domain adaptation must be studied.
• Developing an unsupervised computer-aided approach for recognizing the epileptic onset zone to provide an optimal facility without prior fundamental information of ground truth to attain a clinical decision.
• In attempts to enhance system usability and reduce the training set of epileptic foci localization, data augmentation of focus detection need more attention.
• To further improve epileptic focus identification performance, parameters with high intelligent signal processing and feature extraction methods must be investigated.
• As a new direction in epilepsy clinical diagnosis research, developing robust, efficient, and convenient deep learning systems for localizing epileptic foci is requisite to enhance.