Ecosyste.ms: Awesome

An open API service indexing awesome lists of open source software.

Awesome Lists | Featured Topics | Projects

https://github.com/humancyyborg/cortex_model

A New Approach to Automated Epileptic Diagnosis Using EEG and Probabilistic Neural Network
https://github.com/humancyyborg/cortex_model

biotechnology cancer crytpocurrency epilepsy epilepsy-monitoring epileptic-seizures ethereum

Last synced: 5 days ago
JSON representation

A New Approach to Automated Epileptic Diagnosis Using EEG and Probabilistic Neural Network

Awesome Lists containing this project

README 

        
## EEG Data Analysis and Classification



### Overview

This project involves generating synthetic EEG data, extracting features, and applying a Probabilistic Neural Network (PNN) to classify EEG segments. The primary focus is on distinguishing between different conditions, such as normal, interictal, and ictal states, as well as localizing epileptogenic zones.



The pipeline includes:

- **Generating Synthetic EEG Data**: Simulate EEG data for different conditions.

- **Feature Extraction**: Extract relevant features from the EEG data.

- **Preprocessing**: Normalize the features.

- **Classification**: Apply a Probabilistic Neural Network (PNN) for classification.

- **Evaluation**: Assess model performance using cross-validation and various metrics.



### Getting Started



#### Prerequisites

- Python 3.x

- NumPy

- Pandas

- SciPy

- scikit-learn



You can install the required packages using pip:



```bash

pip install numpy pandas scipy scikit-learn



Generating Synthetic EEG Data

The synthetic EEG data is generated based on three conditions:



Normal: Standard Gaussian noise.

Interictal: Gaussian noise with added sinusoidal oscillations.

Ictal: Higher amplitude Gaussian noise with more pronounced oscillations.

Feature Extraction

Features are extracted using various methods:



Power Spectral Features: Power in different frequency bands.

Fractal Dimensions: Petrosian and Higuchi fractal dimensions.

Hjorth Parameters: Activity, mobility, and complexity.

Statistical Features: Mean, standard deviation, and their absolute values.

Classification

A Probabilistic Neural Network (PNN) is implemented to classify EEG segments. The model is evaluated using 10-fold cross-validation, and performance metrics include accuracy, precision, recall, and F1 score.



Running the Code

To run the experiments and evaluate the model, execute the main.py script:



bash

Copy code

python main.py

Results

The script outputs the results of various classification experiments:



Normal vs Interictal

Normal vs Ictal

Interictal vs Ictal

Epileptogenic zone vs Opposite hemisphere

It displays accuracy, precision, recall, F1 score, and average classification time.



Example Output

plaintext

Copy code

Experiment Results:

------------------

Normal vs Interictal:

  Accuracy: 99.500

  Precision: 99.500

  Recall: 99.500

  F1 Score: 99.500

  Avg. Classification Time: 0.01 seconds



Normal vs Ictal:

  Accuracy: 98.300

  Precision: 98.300

  Recall: 98.300

  F1 Score: 98.300

  Avg. Classification Time: 0.01 seconds



Interictal vs Ictal:

  Accuracy: 96.700

  Precision: 96.700

  Recall: 96.700

  F1 Score: 96.700

  Avg. Classification Time: 0.01 seconds



Epileptogenic zone vs Opposite hemisphere:

  Accuracy: 77.500

  Precision: 77.500

  Recall: 77.500

  F1 Score: 77.500

  Avg. Classification Time: 0.01 seconds

Notes

The current implementation uses synthetic data. For real applications, replace the data generation step with actual EEG data acquisition.

The choice of PNN parameters (e.g., spread) can be tuned for better performance.

Contributing

Feel free to submit issues or pull requests if you have suggestions or improvements. Contributions are welcome!



License

This project is licensed under the MIT License. See the LICENSE file for details.



vbnet