Development and validation of retrospective spinal cord motion time-course estimates (RESPITE) for spin-echo spinal fMRI: Improved sensitivity and specificity by means of a motion-compensating general linear model analysis

• Published in: NeuroImage (Orlando, Fla.) Vol. 44; no. 2; pp. 421 - 427
• Main Authors: Figley, Chase R., Stroman, Patrick W.
• Format: Journal Article
• Language: English
• Published: United States Elsevier Inc 15.01.2009; Elsevier Limited
• Subjects: Algorithms; Analysis; Brain; Cervical Vertebrae; Computer Simulation; Echo-Planar Imaging - methods; Estimates; Evoked Potentials - physiology; fMRI; General linear model (GLM); Human; Humans; Image Enhancement - methods; Image Interpretation, Computer-Assisted - methods; Linear Models; Magnetic Resonance Imaging - methods; Methods; Models, Neurological; Motion; Noise; Physiological motion; Reproducibility of Results; Retrospective motion correction; Retrospective Studies; Sensitivity and Specificity; Spinal cord; Spinal Cord - physiology; Studies; Vertebrae; Human; fMRI; General linear model (GLM); Spinal cord; Analysis; Retrospective motion correction; Physiological motion
• ISSN: 1053-8119; 1095-9572; 1095-9572
• DOI: 10.1016/j.neuroimage.2008.08.040

Cervical spinal cord displacements have recently been measured in relation to the cardiac cycle, substantiating that cord motion in this region reduces both the sensitivity and reproducibility of functional magnetic resonance imaging of the spinal cord (spinal fMRI). Given the ubiquitous and complex nature of this motion, cardiac gating alone is not expected to sufficiently remove these errors, whereas current modeling approaches for spin-echo methods are not specific to motion artifacts, potentially eliminating function-related data along with components of motion-related noise. As such, we have developed an alternative approach to spinal cord motion-compensation, using retrospective spinal cord motion time-course estimates (RESPITE) to forecast a small number of physiological noise regressors. These are generated from the principal components of spinal cord motion, as well as subject-specific cardiac data, and are subsequently included in a general linear model (GLM) analysis. With this approach, the components of motion-related signal fluctuation are modeled, along with functionally-relevant signal changes (i.e., those components fitting the stimulus paradigm), to account for the effects of spinal cord and cerebrospinal fluid (CSF) motion in a thorough, yet discerning, manner. By analyzing 100 previously acquired half-Fourier turbo spin-echo (HASTE) spinal fMRI data sets, along with a collection of null-task data, we show that the implementation of RESPITE reduces the occurrence of both type I (false-positive) and type II (false negative) errors, effectively increasing the specificity (5–6%) and sensitivity (15–20%) to neuronal activity.