Banca de DEFESA: BRITMAN SALCEDO PUMACCOLA

Uma banca de DEFESA de MESTRADO foi cadastrada pelo programa.
STUDENT : BRITMAN SALCEDO PUMACCOLA
Date: 11/12/2025
TIME: 15:00
LOCAL: https://conferenciaweb.rnp.br/webconf/francisco-156
TITLE:

Classification of Alzheimer’s Disease and Lewy Body Dementia using Biomarkers on Resting-State Electroencephalogram Signals through Deep Learning

 

PAGES: 77
BIG AREA: Engenharias
AREA: Engenharia Biomédica
SUBÁREA: Bioengenharia
SPECIALTY: Processamento de Sinais Biológicos
SUMMARY:

This Master thesis investigates the application of deep learning techniques to resting-state

electroencephalogram (EEG) signals for the classification of neurologically healthy older

adults (NOLD), patients with Alzheimer’s disease (AD), and patients with Lewy body

dementia (DLB). A dataset comprising 90 participants (30 AD, 30 DLB, and 30 NOLD)

was analyzed, with AD and DLB groups further divided into mild cognitive impairment

(MCI) and dementia (DEM) categories. Preprocessed resting state EEG signals were

decomposed into standard frequency bands (delta, theta, alpha, beta, and gamma), and

normalized power spectral density (PSD) features were extracted from 19 channels. Two

deep learning models — Long Short-Term Memory (LSTM) networks and Transformer-

based architectures — were implemented and validated using the Leave One Subject

Out (LOSO) cross-validation scheme, to simulate real clinical generalization. The LSTM

classifier achieved an average accuracy of 94.97%, with 95.03% sensitivity and 94.96%

specificity across three binary group comparisons (AD vs. DLB, AD vs. NOLD, and DLB vs.

NOLD) and three binary category comparisons (DEM vs. MCI, DEM vs. CTRL, and MCI

vs. CTRL). The Transformer model yielded even higher performance, with 96.40% average

accuracy, 97.29% sensitivity, and 95.56% specificity. In multi-class classification, both

approaches demonstrated robust accuracy: 86.67% (LSTM) and 87.78% (Transformer) for

five-class discrimination (AD-DEM, AD-MCI, DLB-DEM, DLB-MCI, and NOLD-CTRL),

and above 90% for three-class group (AD, DLB, and NOLD) and category (DEM, MCI,

and CTRL) distinctions. These results demonstrate that deep learning applied to resting

state EEG data can effectively distinguish between AD, DLB, MCI, and healthy aging,

providing a non-invasive, accurate, and generalizable approach for early and differential

detection of dementia.

COMMITTEE MEMBERS:
Presidente - Interno ao Programa - 1545987 - FRANCISCO JOSE FRAGA DA SILVA
Membro Titular - Examinador(a) Interno ao Programa - 1672975 - JOAO RICARDO SATO
Membro Titular - Examinador(a) Externo ao Programa - ***.158.528-** - LUCAS REMOALDO TRAMBAIOLLI - Harvard
Membro Suplente - Examinador(a) Interno ao Programa - 1761107 - RICARDO SUYAMA
Membro Suplente - Examinador(a) Externo ao Programa - ***.727.270-** - TIAGO OLIVEIRA WEBER - UFRGS