Loading...

An Investigation of Resting-State Eeg Biomarkers Derived from Graph of Brain Connectivity for Diagnosis of Depressive Disorder

Arabpour, Mohammad Reza | 2020

522 Viewed
  1. Type of Document: M.Sc. Thesis
  2. Language: Farsi
  3. Document No: 53150 (05)
  4. University: Sharif University of Technology
  5. Department: Electrical Engineering
  6. Advisor(s): Hajipour, Sepideh
  7. Abstract:
  8. Among the most costly diseases that affect a person's quality of life throughout his or her life, mental disorders (excluding sleep disorders) affect up to 25 percent of people in any community. One of the most common types of these disorders in Iran is depressive disorder, which according to official statistics, 13% of Iranians have some symptoms of it. Until now, the diagnosis of this disease has been traditionally done in clinics with interviews and questionnaires tests based on behavioral psychology and using symptom assessment. Therefore, there is a relatively low accuracy in the treatment process. Nowadays, with the help of functional brain imaging such as electroencephalogram (EEG) along with signal and image processing tools, features called biomarkers can be extracted in psychology and psychiatric clinics to be used to help distinguish between people with the disorder and normal people. Among all biomarkers, or signal characteristics associated with the disease, our main focus in this study is on brain connections; Because, according to the results of previous research, brain connections can better reflect the functional characteristics of the brain and mental states such as depression than other behavioral methods and even other biomarkers. In this dissertation, we will first provide an introduction to the issue and then an overview of the scientific literature available in clinics and written articles. And then we will present a report of classification using classical machine learning on a set of data with two labels, depressed and healthy. This data is recorded in the Atieh Psychology Clinic . Finally, we present our proposed method, which is based on classification using pathway graph kernels, in three types: fully rotational (normal path kernel), semi-rotating and non-rotating. The results obtained from applying the proposed method to the data set show that the proposed method has been able to improve in some cases, compared to the classical classification method
  9. Keywords:
  10. Biomarker ; Electroencphalogram ; Brain Connectivity ; Resting State Brain ; Functional Brain Imaging ; Depressive Disorder

 Digital Object List

 Bookmark

No TOC