Awesome Open Source
Awesome Open Source

Diagnosis ADHD with Convolutional-LSTM Model

Motivation

  • Diagnosing mental disorders is a considerably complex task for behavioral health professionals
  • Many factors complicate the process:
    • People exhibit individual behaviors with the symptoms of their disorder(s)
    • No objective biological markers associated with mental disorders
    • Mental disorders often overlap with one to many others
    • Similarity of symptoms among different diseases can lead to inaccurate diagnosis 
  • Potential Solution lies with functional Magnetic Resonance Imaging (fMRI) technology

functional Magnetic Resonance Imaging

  • Measures brain activity by detecting changes associated with blood flow
    • Known as blood-oxygen-level dependent (BOLD) method
  • This technique relies on the fact that blood flow and neuronal activation are coupled.
  • When an area of the brain is in use, blood flow to that region also increases
  • Does so to provide energy to the neurons, which do not have internal reserves of energy.
  • fMRI machine captures blood flow and “lights up” brain areas in images
  • Indicates that part of the brain is responsible for handling a certain activity

Objective

  • Construct a hybrid model that captures and analyzes both the spatial and temporal aspects of an fMRI dataset
  • This model will consist of a Convolutional Neural Network and Recurrent Neural Network
  • Convolutional Neural Network:
    • Retrieves spatial features of the data
    • Extracts the details of active areas in the brain
  • Recurrent Neural Network
    • Retrieves the temporal features of the data
    • Model the flow of the blood that is associated to certain disorders (or activities)
  • Goal: Construct a 3D Convolutional Neural Network + LSTM-based RNN to diagnose the ADHD disorder with the provided fMRI data sample

Currently In Progress

Related Awesome Lists
Top Programming Languages
Top Projects

Get A Weekly Email With Trending Projects For These Topics
No Spam. Unsubscribe easily at any time.
Jupyter Notebook (157,682
Deep Learning (36,991
Tensorflow (22,458
Neural (16,689
Convolutional Neural Networks (12,568
Flow (10,535
Keras (10,471
Lstm (5,539
Medical Imaging (740
Neuroimaging (624
Fmri (205
Convlstm (26