Nombre: LORRAINE MARQUES ALVES
Fecha de publicación: 23/03/2023
Supervisor:
Nombre |
Papel |
|---|---|
| PATRICK MARQUES CIARELLI | Advisor * |
Junta de examinadores:
| Nombre |
Papel |
|---|---|
| ELIAS SILVA DE OLIVEIRA | External Examiner * |
| JUGURTA ROSA MONTALVÃO FILHO | External Examiner * |
| KARIN SATIE KOMATI | External Examiner * |
| PATRICK MARQUES CIARELLI | Advisor * |
| RODRIGO VAREJÃO ANDREÃO | External Examiner * |
Sumario: Electroencephalogram (EEG) is an important source of signals to support diagnosis of mental disorders and, every day more research is looking for ways to use these signals to improve the quality of medical diagnosis. Diseases such as Schizophrenia, Epilepsy, Attention Deficit Hyperactivity Disorder (ADHD), Depression, Alcoholism, among others, are examples WHERE the application of reading and decoding EEG techniques are of great value in supporting medical diagnosis. EEG revealed in the dynamic brain studies that global neural activity can be described by a limited number of electrical potential topographic maps of scalp, called microstates. In this work, two new methods are proposed for decoding EEG signals to application in the detection of Schizophrenia, Depression and ADHD exploring approaches applied to the Electroencephalogram (EEG) microstates. One proposed method is based on graph theory and the other on natural language processing. Both proposed methods allow an understanding of how the brain signals of an individual with mental disorder differ from healthy control. Graph theory allowed the determination of important topological characteristics from the constructed microstate networks, resulting in an average accuracy of 100.0% in schizophrenia, 100.0% in depression and in ADHD the average accuracy was 99.44% (ADHD vs Healthy Control) and 98.61% (ADD vs ADHD-C vs Healthy Control). The application of natural language processing on symbolic sequences of microstates revealed the importance of contextual information from the semantic point of view in the characterization of patients with mental disorders, resulting in an average accuracy of 100.0% in schizophrenia,
98.47% in depression and in ADHD the average accuracy was 99.38% (ADHD vs Healthy Control) and 98.19% (ADD vs ADHD-C vs Healthy Control). These proposed methods contribute to the automation of mental disorder diagnosis, being a promising tool in the development of psychiatry based on artificial intelligence.
Key-words: Mental Disorders. EEG. Microstates. Graph Theory. Natural Language Processing.
