Please use this identifier to cite or link to this item:
http://hdl.handle.net/10603/446764
Title: | Detection of Brain Diseases Using Reconstructed Phase Space Portraits and Recurrence Plots of Electroencephalogram |
Researcher: | Ilakiyaselvan, N |
Guide(s): | Nayeemulla Khan |
Keywords: | Computer Science Computer Science Interdisciplinary Applications Engineering and Technology |
University: | Vellore Institute of Technology (VIT) University |
Completed Date: | 2022 |
Abstract: | Brain abnormalities affect the functions of the brain in people of all ages and from all socio-economic backgrounds. These abnormalities of the brain are observable through the electroencephalogram (EEG) signals that, measures the brain s electrical activity. Methods to automatically identify brain diseases from the easily recordable scalp EEG signals would greatly assist neurophysiologists in patient diagnosis. Automatic detection of brain diseases has been attempted in earlier studies using the spatial and temporal characteristics of the EEG employing signal processing, image processing and machine learning techniques. However accurate detection across multiple brain diseases is a far more challenging problem given the dynamic nature of the EEG signal. newlineEEG signal is nonlinear and chaotic in nature. We employ Reconstructed Phase Space (RPS) portraits and Recurrence plots to model these nonlinear dynamics. The main focus is to utilize these portraits as a mechanism to represent dynamic systems, while also exploiting the two-dimensional image representation as a feature representation mechanism. In this study, we examine the performance of CNN-based deep learning models as feature extractors to assist in improved disease screening of EEG signals by classifying them as normal or diseased using RPS images and recurrence plots. newlineThe novel use of RPS, RP and CNNs for disease identification from short segments (2-6 sec) of the relevant EEG signal is proposed. The CNN models result in high accuracy exceeding that reported in literature for disease identification and classification into normal/diseased cases. An ensemble voting classifier further improves the accuracy in disease identification. newlineThe modeling of non-linear dynamics in the EEG and the use of CNNs for learning salient features suitable for accurately identifying the brain disease in an end-to-end system is a novel contribution of this study. The high performance of the system makes it feasible to study its utility as a diagnostic support tool. newline |
Pagination: | i-xiv,144 |
URI: | http://hdl.handle.net/10603/446764 |
Appears in Departments: | School of Computing Science and Engineering VIT-Chennai |
Files in This Item:
File | Description | Size | Format | |
---|---|---|---|---|
01_title.pdf | Attached File | 49.52 kB | Adobe PDF | View/Open |
02_prelim pages.pdf | 551.25 kB | Adobe PDF | View/Open | |
03_contents.pdf | 46.87 kB | Adobe PDF | View/Open | |
04_abstract.pdf | 163.1 kB | Adobe PDF | View/Open | |
05_chapter_1.pdf | 547.27 kB | Adobe PDF | View/Open | |
06_chapter_2.pdf | 560.89 kB | Adobe PDF | View/Open | |
07_chapter_3.pdf | 130.97 kB | Adobe PDF | View/Open | |
08_chapter_4.pdf | 1.07 MB | Adobe PDF | View/Open | |
09_chapter_5.pdf | 942.94 kB | Adobe PDF | View/Open | |
10_chapter_6.pdf | 1.04 MB | Adobe PDF | View/Open | |
11_chapter_7.pdf | 2.93 MB | Adobe PDF | View/Open | |
12_chapter_8.pdf | 47.64 kB | Adobe PDF | View/Open | |
13_annexures.pdf | 247.84 kB | Adobe PDF | View/Open | |
80_recommendation.pdf | 98.03 kB | Adobe PDF | View/Open |
Items in Shodhganga are licensed under Creative Commons Licence Attribution-NonCommercial-ShareAlike 4.0 International (CC BY-NC-SA 4.0).
Altmetric Badge: