Brain Vectors by Vecteezy
Simple Convolutional Neural Network intended to diagnose fictitious MRI images, labeling new inputs as one of the following classes:
No Impairment, Very Mild Impairment,
Mild Impairment, Moderate Impairment
· Report Bug
· Request Feature
Table of Contents
In a deeply personal quest to combat a devastating and pervasive disease, I embarked on a journey to develop a neural network model to read fictitious MRI data and detect the early signs of Alzheimer's disease. This endeavor is motivated by a profound family history that has been haunted by Alzheimer's for generations. The emotional toll of Alzheimer's, coupled with the urgent need for early diagnosis and intervention, fueled my determination to make a difference.
My journey began by delving into the world of medical imaging and artificial intelligence. I gathered fictitious MRI data, sourced from kaggle.com, which mirrors the complexities of real-world medical images, to construct a neural network model. This model has been meticulously trained to analyze subtle patterns and anomalies within the brain, with a specific focus on identifying the early indicators of Alzheimer's disease.
My hope is that this project will serve as a beacon of hope for others facing Alzheimer's, a testament to the power of technology, and a tribute to the loved ones who have inspired it. Together, we can shine a light on this dark path and take meaningful steps toward early diagnosis, treatment, and ultimately, a cure for Alzheimer's disease.
Below are the major Python frameworks used for this project
Installing this software is as simple as the few steps below:
First, pull this git repo to a local directory and cd into that directory.
Next, upgrade pip:
python -m pip install --upgrade pip
Then pip install the included requirements.txt file.
pip install -r requirements.txt
Now you are ready to run tests or the front-end application.
This software utility offers a simple and accessible way to predict Alzheimer's disease progression using MRI data. Here's how to use it:
Starting the Web User Interface (UI):
./start_front_end.sh
Sending MRI Files:
To predict Alzheimer's disease progression, send MRI image files to the http://127.0.0.1:5000/predict endpoint of the Flask server. Note these MRI files should be one of the fictitious MRIs present in the data/test/ directory. This model has zero ability to predict actual MRI images.
Interpreting Predictions:The utility will return a prediction for the diagnosis, categorizing it into one of four levels of impairment:
- No Impairment
- Very Mild Impairment
- Mild Impairment
- Moderate Impairment
Using the Web UI:
Accessing the Web Interface: Open your web browser and navigate to http://127.0.0.1:5000. This will take you to a basic Graphical User Interface (GUI).
Selecting Impairment Categories:
On the web UI, you will find four buttons, each corresponding to a different level of impairment. By clicking on one of these buttons, you can initiate a prediction for that specific impairment category.
Testing the Model:
The utility will randomly select an MRI image from the training set for the chosen impairment category and run it through the predictive model.
This utility provides a convenient and user-friendly way to predict Alzheimer's disease progression, making it accessible to both professionals and non-experts. By following these simple steps, you can quickly assess the likelihood of Alzheimer's disease in MRI images and gain insights into its progression.
Distributed under the MIT License. See LICENSE.txt for more information.
Jonathan Trull - jttrull0@gmail.com
Project Link: https://github.com/jtrull101/alz-mri-neural-network
