Bairy, Muralidhar G and LIh, Oh Shu and Hagiwara, Yuki and Puthankattil, Sudha D and Faust, Oliver and Niranjana, S and Acharya, Rajendra U Automated Diagnosis of Depression Electroencephalograph Signals Using Linear Prediction Coding and Higher Order Spectra Features. Journal of Medical Imaging and Health Informatics, 7 (8). pp. 1-6. ISSN 2156-7026
![]() |
PDF
3567.pdf - Published Version Restricted to Registered users only Download (1MB) | Request a copy |
Abstract
Depression is a mental disorder that negatively affects the day to day activities of a patient. Diagnosing depression is of paramount importance to reduce suffering for the patient and support network. Electroencephalograph (EEG) signal variations can indicate neurological diseases associated with mental trauma. EEG being a non-invasive technique, is widely used to analyse various brain disorders. However, to detect and interpret the minute signal changes a computer-aided diagnosis (CAD) system is developed. Higher order statistic based parameters, such as variance, kurtosis, normalized kurtosis, skewness, normalized skewness is extracted from the linear predictive coding (LPC) residuals. Seven different feature ranking methods are used to test and rank the extracted features. Feature ranking using Receiver Operating Characteristic (ROC) gave the best classification accuracy of 94.30%, the sensitivity of 91.46% and specificity of 97.45% using a bag tree classifier
Item Type: | Article |
---|---|
Uncontrolled Keywords: | Bagged Tree Classification, Depression, EEG Signals, Feature Ranking Methods, Higher Order Statistics, Kurtosis, Linear Predictive Coding Residual, Skewness, Receiver Operating Characteristics, Variance |
Subjects: | Engineering > MIT Manipal > Biomedical |
Depositing User: | MIT Library |
Date Deposited: | 13 Dec 2017 09:12 |
Last Modified: | 13 Dec 2017 09:12 |
URI: | http://eprints.manipal.edu/id/eprint/150227 |
Actions (login required)
![]() |
View Item |