Utilizing Deep Learning on Resting State fMRI Data to Predict Memory Outcome following Temporal Lobe Resection for Intractable Epilepsy
- 0 Collaborators
The idea of this project is to harness the power of deep learning tools to find patterns and build a classifier the resting-state fMRI data of patients with Temporal Lobe resection for intractable epilepsy. ...learn more
Project status: Concept
Intel Technologies
DevCloud,
Intel Opt ML/DL Framework,
Intel Python
Overview / Usage
Data: The data was collected at the Epilepsy Care Center, Department of Neurology, Thomas Jefferson University Hospital following the temporal lobe resection. The data are the resting state fMRI of these patients. To observe the default brain networks, fMRI of whole brain is taken at very short interval for 120 times, creating a 3D - time series data (4D data) for each patient.
We are exploring the idea of finding some kind of predictive power in resting state fMRI, by building a classifier on the data set mentioned above. This is a challenging task as using machine learning on 4D data is still very new. Thus, our initially task have been finding out the right type of architecture for this problem, and designing experiments. In next few months, we will be carrying out experiments using Intel DevCloud.