Skip to content

Latest commit

 

History

History
32 lines (22 loc) · 2.44 KB

README.md

File metadata and controls

32 lines (22 loc) · 2.44 KB
Raw

ADHDeepNet: ADHD Diagnosis Using EEG Data and Deep Learning

Introduction

ADHDeepNet employs deep learning techniques and EEG signals to diagnose Attention Deficit Hyperactivity Disorder (ADHD). It integrates Convolutional Neural Networks with elements from EEGNet, Inception/Xception, and Squeeze and Excitation networks.

Model Architecture

Inspired by EEGNet, Inception, Squeeze and Excitation, and Xception models, ADHDeepNet is specifically designed to process raw EEG signals efficiently for ADHD classification.

Key Features

  • Efficient Processing: Directly operates on raw EEG signals, minimizing preprocessing needs.
  • Tailored for ADHD: Architectural modifications geared towards ADHD classification.
  • Hybrid Architecture: Combines various proven models for enhanced performance.
  • Ablation Study: Conducted to ensure each module and block is essential to the model architecture, validating the design choices.

Data Augmentation

Due to challenges in EEG data collection, ADHDeepNet leverages Data Augmentation (DA) to increase the dataset's size and diversity, essential for EEG signal variability.

Gaussian Noise Addition

  • Objective: To improve model robustness by learning more discriminative features.
  • Method: Perturbing raw EEG signals with random noise from a Gaussian distribution, encouraging the model to learn general patterns rather than specific training data traits.
  • Benefit: Especially effective for EEG data, aiding the model in generalizing across diverse EEG recordings for ADHD diagnosis.

Evaluation Methodology

  • Validation Technique: Employs (10-2)-fold cross-subject validation.
  • Hyperparameter Optimization: Uses Bayesian Optimization for model tuning to achieve optimal performance in ADHD diagnosis.

Code and Usage

The code for ADHDeepNet is provided as a Jupyter Notebook and can be accessed at the following link: ADHDeepNet Jupyter Notebook. This notebook includes detailed instructions for setting up the environment, loading the dataset, and running the model.

The raw EEG dataset used for training and evaluating ADHDeepNet is publicly available and can be downloaded from: EEG Data for ADHD and Control Children. Please refer to this link for access to the dataset and its documentation.