Quantitative analysis tool for clinical functional MRI in mild traumatic brain injury

Abstract

Functional magnetic resonance imaging (fMRI) has been available commercially for clinical diagnostic use for many years. However, both clinical interpretation of fMRI by a neuroradiologist and quantitative analysis of fMRI data can require significant personnel resources that exceed reimbursement. In this report, a fully automated computer-based quantification methodology (Enumerated Auditory Response, EAR) has been developed to provide an auditory fMRI assessment of patients who have suffered a mild traumatic brain injury. Fifty-five study participants with interpretable auditory fMRI sequence data were assessed by EAR analysis, as well as both clinical radiologist fMRI interpretation and voxelwise general linear model (GLM) analysis. Comparison between the clinical interpretation and the two computer analysis methods resulted in 67% concordance (identical), 32% nearconcordance (one level difference), and 1% discordant. Comparison between the clinical computer-based quantification (EAR) and GLM analysis yielded significant correlations in right and left ear responses (p⟨0.05) for the full subject group. Automated fMRI quantification analysis equivalent to EAR might be appropriate for both future research projects with constrained resources, as well as possible routine clinical use.