Meeting Banner
Abstract #4160

A Novel Method for Robust Automated Thresholding in Pre-surgical fMRI using a Single Functional Run.

Tynan Stevens 1,2 , David Clarke 3,4 , Ryan D'Arcy 5,6 , Gerhard Stroink 1 , and Steven Beyea 2,7

1 Physics, Dalhousie University, Halifax, NS, Canada, 2 Neuroimaging Research Lab, BIOTIC, Halifax, NS, Canada, 3 Surgery, Dalhousie University, Halifax, NS, Canada, 4 Neurosurgery, QEII Health Sciences Centre, Halifax, NS, Canada, 5 Applied Science, Simon Frasier University, Burnaby, BC, Canada, 6 Surrey Memorial Hospital, Surrey, BC, Canada, 7 Radiology, Dalhousie University, Halifax, NS, Canada

We demonstrate a novel data-driven method for selecting thresholds for pre-surgical fMRI data, based on reliability of the activation patterns in just a single fMRI run. Our new method incorporates spatial information not present in histogram based thresholding methods, and alleviates the need for test-retest imaging of existing reliability optimization methods. The new method produces significantly higher test-retest overlap when compared to established threshold optimization methods, particularly for low CNR situations like language mapping in patient populations. This analysis therefore provides the most robust automated thresholds, and unlike other techniques can be applied to any existing fMRI paradigm without modification.

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

Join Here