🧠 ThinkTracer: fMRI Brain Task Classifier

A simulated brain activity classification pipeline using synthetic fMRI BOLD signals. The project generates realistic voxel activation patterns across different cognitive tasks (e.g., memory, language, motor) and classifies them using PCA + SVM. It is deployed using Streamlit Cloud for interactive use.

🚀 Live Demo

👉 Launch the Streamlit App (https://thinktracer.streamlit.app/)

📂 Project Structure

bash 📁 fMRI-Brain-Task-Classifier/

├── app.py # Streamlit application script

├── pca_model.pkl # Trained PCA dimensionality reduction model

├── task_classifier.pkl # Trained SVM classifier

├── requirements.txt # Dependencies for Streamlit Cloud

├── fMRI_task_classifier.ipynb # Notebook for data simulation and model training

└── README.md # Project description

🧪 Project Overview

This project simulates and classifies fMRI brain data by:

Simulating voxel-level BOLD activity for tasks like:

Memory recall

Language comprehension

Motor function

Introducing subject variability and an age factor

Applying PCA for dimensionality reduction

Training a Logistic regression model for baseline testing

Training an SVM model to classify tasks based on reduced features

Deploying the model via Streamlit Cloud

📊 Dataset Details

The synthetic dataset simulates multiple aspects of real fMRI data:

🧬 Voxel-wise activity for each task

👤 Subject variability to simulate inter-individual differences

👵 Age factor to reflect cognitive and signal changes across age

🔉 Noise to emulate scanner and neural variability

Each row represents one subject-task combination.

🧠 Model Workflow

Generate data with realistic patterns for each task

Reduce dimensionality using PCA

Train classifier (SVM) on PCA outputs

Deploy with Streamlit for easy user input + predictions

📈 How to Use

Upload a CSV file with fMRI features (simulated or real)

The app will:

Transform it using PCA

Predict the cognitive task

View the prediction results live in the app

🛠 Requirements

nginx

streamlit numpy pandas scikit-learn joblib

Install locally with:

bash

pip install -r requirements.txt

📚 Technologies Used

Python

Scikit-learn

PCA

SVM

Streamlit

🧑‍💻 Author

Shrutaswini [github.com/Shrutaswini]

Feel free to connect on LinkedIn

💡 Future Ideas

Add task difficulty modeling

Test with real open-source fMRI datasets (e.g., HCP, OpenNeuro)

Include visualization of voxel activations