Meeting Banner
Abstract #0243

Fully Automated Learning based Method for resting state fMRI Connectomics Analysis

Arathi Sreekumari1, Radhika Madhavan1, Rakesh Mullick1, Teena Shetty2, Pratik Mukherjee3, Joseph Masdeu4, Luca Marinelli5, and Suresh Emmanuel Joel1

1GE Global Research, Bangalore, India, 2Hospital for Special Surgery, New York, NY, United States, 3University of California, San Francisco, San Francisco, CA, United States, 4Houston Methodist, Houston, TX, Houston, TX, United States, 5GE Global Research, Niskayuna, NY, United States

Machine learning approaches are increasingly being used to identify discriminative features derived from functional connectome data that best separate a diseased group from healthy cohorts. Here, we propose a novel framework for longitudinal prediction of disease outcome, using a combination of unsupervised and supervised learning approaches. Using this framework, we achieve 81% accuracy for prediction of mild traumatic brain injury outcome at 3-months by learning features from functional connectomes at the acute stage of injury (<1 week).

This abstract and the presentation materials are available to members only; a login is required.

Join Here