MindSyncPlus

MindSyncPlus is a web-based application designed to aid in detecting the stage of Alzheimer’s disease from MRI scans. The system integrates custom-built deep learning models with an intuitive user interface to provide accurate, stage-specific diagnoses and insights into Alzheimer’s progression.

Table of Contents

• Features
• Technologies Used
• Installation
• Usage
• Model Details
• Model Architectures
• Model Performance
• Screenshots and Video
• Future Enhancements
• Contributing
• License

Features

• Accurate Diagnosis: Achieves 85%+ accuracy in identifying Alzheimer’s stages from MRI scans.
• User-Friendly Interface: Simplifies the process for uploading scans and obtaining predictions.
• Data Visualization: Provides visual insights for better understanding of model predictions.
• Modular Design: Built with scalability in mind to accommodate future enhancements.

Technologies Used

• Programming Language: Python
• Frameworks and Libraries:
  • Backend: Flask
  • Frontend: HTML5, CSS3, JavaScript, Bootstrap
  • Machine Learning: TensorFlow, NumPy, Pandas, Scikit-learn
• Database: MySQL
• Developer Tools: VS Code, Jupyter Notebooks, Google Colab

Installation

1. Clone the repository:

   git clone https://github.com/Angad-2002/MindSyncPlus.git

2. Navigate to the project directory:

   cd MindSyncPlus

3. Create a virtual environment (optional):

   python -m venv venv
   source venv/bin/activate  # For Linux/macOS
   venv\Scripts\activate    # For Windows

4. Install the required dependencies:

   pip install -r requirements.txt

5. Configure the database:

   • Create a MySQL database.
   • Update the database credentials in the config.py file.

6. Run the application:

   python app.py

7. Access the application in your browser at http://127.0.0.1:5000.

Usage

1. Upload MRI Scans: Navigate to the upload section and select the MRI scans you want to analyze.
2. Analyze Results: The system will predict the stage of Alzheimer’s and display the results.
3. Visual Insights: Explore detailed visualizations of the analyzed data.
4. Manage Records: Use the database to store and retrieve past analyses.

Model Details

The application uses three different deep learning models to analyze MRI scans:

1. Inception V3:

   • Pre-trained on the ImageNet dataset.
   • Fine-tuned to classify Alzheimer’s stages.
   • Known for its efficiency and accuracy in image classification tasks.

2. VGG19:

   • Another pre-trained model from the ImageNet family.
   • Fine-tuned for Alzheimer’s stage detection.
   • Excels in handling deep hierarchical feature representations.

3. Custom Model:

   • A custom-built convolutional neural network (CNN) tailored for this project.
   • Designed with layers optimized for detecting patterns specific to Alzheimer’s MRI scans.
   • Balances performance and computational efficiency for deployment.

Model Architectures

1. Inception V3

• Architecture:
  • Consists of inception modules that use convolutional filters of different sizes in parallel.
  • Includes auxiliary classifiers for better gradient flow.
  • Employs factorized convolutions to reduce computational cost.

2. VGG19

• Architecture:
  • Contains 19 layers: 16 convolutional layers and 3 fully connected layers.
  • Uses 3x3 filters with stride 1 and same padding.
  • Focuses on deep and uniform architecture with a fixed convolutional window.

3. Custom CNN Model

• Architecture:
  • Input Layer: Accepts MRI scans in a standardized format.
  • Convolutional Layer 1: 64 filters, 3x3 kernel, ReLU activation.
  • MaxPooling Layer 1: 2x2 pooling window.
  • Convolutional Layer 2: 64 filters, 3x3 kernel, ReLU activation.
  • MaxPooling Layer 2: 2x2 pooling window.
  • Flatten Layer: Converts 2D matrix to 1D vector.
  • Dense Layer 1: 128 units, ReLU activation.
  • Dropout Layer: 50% dropout to reduce overfitting.
  • Output Layer: 4 units with softmax activation for multi-class classification.

Diagram

Input Image
    |
Conv2D (64 filters, 3x3) -> ReLU -> MaxPooling (2x2)
    |
Conv2D (64 filters, 3x3) -> ReLU -> MaxPooling (2x2)
    |
Flatten
    |
Dense (128 units) -> ReLU -> Dropout (50%)
    |
Dense (4 units) -> Softmax

Model Performance

• Accuracy: 85%+
• Metrics: Evaluated using precision, recall, F1-score, and confusion matrix.
• Dataset: Trained on publicly available Alzheimer’s MRI datasets, with preprocessing to ensure data quality.

Screenshots and Video

Screenshots

1. Home Page:

   

2. Upload Section:

   

3. Sign-Up/Login Page:

   

4. Results Page:

   

Video Demo

A video walkthrough demonstrating the application’s functionality is available.

Future Enhancements

• Integrate additional deep learning models to improve accuracy.
• Support for other neurodegenerative diseases.
• Implement role-based access control for enhanced security.
• Add cloud storage support for MRI scans and results.

Contributing

Contributions are welcome! To contribute:

1. Fork this repository.
2. Create a new branch:

   git checkout -b feature-name

3. Commit your changes:

   git commit -m "Add feature"

4. Push to the branch:

   git push origin feature-name

5. Create a pull request.

Contributors

• Ashish GitHub
• Siddharth GitHub
• Angad Singh GitHub

License

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

---

Maintainer: Angad Singh
GitHub | LinkedIn