Graph Fourier Transform

This repository demonstrates a workflow for analyzing EEG data using Graph Fourier Transform (GFT). Below are the steps to reproduce the results.

---

Step 1 - Downloading the Dataset

1. Download the dataset from OpenNeuro.
   • Dataset used: EEG dataset for brain activity analysis.
   • Save the dataset in the folder: data/Data-OpenNeuro.

---

Step 2 - Data Preprocessing

1. Run the preprocessing script to convert raw EEG data into a usable format:

   run src/data_preprocessing/load_save_data.m
   

2. This script processes the data and saves the output in the folder: data/ProcessedMATFiles

---

Step 3 - Constructing the Graph

1. If you wish to corroborate, construct a graph representation of the EEG data using the following script:

run src/graph_construction/graph_construction.m

2. Pass the .mat file from data/ProcessedMATFiles as the parameter to generate the graph adjacency matrix and Laplacian.

---

Step 4 - Feature Extraction

1. Extract relevant features from the constructed graph using:

 run src/feature_extraction/extract_features.m

This script processes all the files in data/ProcessedMATFiles and generates a feature matrix containing the computed features for all EEG samples.

2. It also generates a labels array categorizing each sample into groups such as Alzheimer’s, Frontal Dementia, or Healthy controls.

---

Step 5 - Classification

1. Run machine learning classification models using:

    run src/model/run_classification.m

This script:

• Splits the data into training and testing sets.
• Evaluates classifiers including k-NN, Random Forest, Logistic Regression, SVM, Naive Bayes, Neural Networks, and more.
• Computes performance metrics such as accuracy, precision, recall, and F1-score for each model.
• Prints the results in the MATLAB console.

---

Step 6 - Plotting the Results

1. Generate visualizations to interpret the results using the plotting scripts:

run src/Plotting/<script_name>.m

2. Plots are saved in a designated folder, such as:

results/plots

---

Folder Structure

project/
├── data/
│   ├── Data-OpenNeuro/           % Raw EEG data
│   ├── ProcessedMATFiles/        % Preprocessed EEG data
├── src/
│   ├── data_preprocessing/       % Scripts for preprocessing
│   ├── graph_construction/       % Scripts for graph construction
│   ├── feature_extraction/       % Scripts for extracting features
│   ├── model/                    % Scripts for training classifiers
│   ├── Plotting/                 % Scripts for generating plots
├── results/
│   ├── plots/                    % Folder to save generated plots
├── README.md                     % This file

This structured workflow allows for seamless analysis of EEG data using GFT. For any issues, please consult the respective scripts and their comments.