Klasifikasi Kondisi Mental Berdasarkan Brain Connectivity pada Sinyal EEG Menggunakan CNN

Taufan, Andika Dwi Muhammad (2026) Klasifikasi Kondisi Mental Berdasarkan Brain Connectivity pada Sinyal EEG Menggunakan CNN. Other thesis, Institut Teknologi Sepuluh Nopember.

Text 5024201016-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (7MB) | Request a copy

Abstract

Electroencephalography (EEG) merupakan metode perekaman aktivitas listrik otak yang dapat digunakan untuk mengidentifikasi berbagai kondisi mental manusia. Salah satu pendekatan untuk menganalisis interaksi antarwilayah otak adalah Brain Connectivity. Penelitian ini bertujuan untuk mengklasifikasikan kondisi mental, yaitu concentrating, relaxed, dan neutral, berdasarkan fitur Brain Connectivity dari sinyal EEG menggunakan metode Convolutional Neural Network (CNN). Data EEG yang digunakan berasal dari empat kanal, yaitu AF7, AF8, TP9, dan TP10. Tahapan pengolahan data meliputi filterisasi sinyal ke dalam lima subband frekuensi EEG, yaitu delta, theta, alpha, beta, dan gamma; segmentasi sinyal; serta ekstraksi fitur Brain Connectivity menggunakan metode Pearson Correlation dan Coherence. Fitur yang diperoleh direpresentasikan dalam bentuk matriks konektivitas berukuran 4 × 4 dan digunakan sebagai masukan model CNN. Penelitian ini menggunakan 245 sampel data yang dibagi menjadi data latih dan data uji dengan rasio 80:20. Tiga model CNN dikembangkan dan dievaluasi menggunakan metrik accuracy, precision, recall, dan F1-score. Hasil pengujian menunjukkan bahwa Model 1 memperoleh performa terbaik dengan accuracy sebesar 83,67%, macro precision sebesar 0,83, macro recall sebesar 0,84, dan macro F1-score sebesar 0,83. Hasil penelitian menunjukkan bahwa kombinasi fitur Brain Connectivity berbasis Pearson Correlation dan Coherence dengan CNN mampu mengklasifikasikan kondisi mental dari sinyal EEG dengan performa yang baik.
=====================================================================================================================================
Electroencephalography (EEG) is a method for recording the brain’s electrical activity and can be used to identify various human mental conditions. One approach for analyzing interactions among brain regions is Brain Connectivity. This study aims to classify mental conditions, namely concentrating, relaxed, and neutral states, based on Brain Connectivity features extracted from EEG signals using a Convolutional Neural Network (CNN). The EEG data were obtained from four channels: AF7, AF8, TP9, and TP10. The data processing stages included signal filtering into five EEG frequency subbands, namely delta, theta, alpha, beta, and gamma, signal segmentation, and Brain Connectivity feature extraction using Pearson Correlation and Coherence. The extracted features were represented as 4 × 4 connectivity matrices and used as inputs to the CNN model. A total of 245 samples were used in this study and divided into training and testing sets using an 80:20 ratio. Three CNN models were developed and evaluated using accuracy, precision, recall, and F1-score metrics. The experimental results showed that Model 1 achieved the best performance with an accuracy of 83.67%, a macro precision of 0.83, a macro recall of 0.84, and a macro F1-score of 0.83. The results demonstrate that the combination of Brain Connectivity features based on Pearson Correlation and Coherence with CNN can classify mental conditions from EEG signals with good performance.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	EEG, Brain Connectivity, Pearson Correlation, Coherence, CNN, Mental State.
Subjects:	H Social Sciences > HA Statistics > HA31.3 Regression. Correlation. Logistic regression analysis. Q Science > QA Mathematics > QA76.87 Neural networks (Computer Science) Q Science > QP Physiology > Q376.5 Electroencephalography (EEG)
Divisions:	Faculty of Intelligent Electrical and Informatics Technology (ELECTICS) > Computer Engineering > 90243-(S1) Undergraduate Thesis
Depositing User:	Andika Dwi Muhammad Taufan
Date Deposited:	24 Jul 2026 02:00
Last Modified:	24 Jul 2026 02:02
URI:	http://repository.its.ac.id/id/eprint/136585

Actions (login required)

View Item