A novel multi-echo fMRI weighting strategy using principal component analysis for BOLD contrast sensitivity enhancement

Introduction: In functional magnetic resonance imaging (fMRI), acquisition parameters must balance the tradeoffs between spatial coverage and resolution, temporal resolution, and signal contrast. Blood oxygenation level dependent (BOLD) fMRI studies commonly employ a 2D multi-slice T2*-weighted echo-planar imaging (EPI) acquisition to capture changes in T2* decay due to BOLD signal fluctuation, with a single K-space traversal per slice, per radiofrequency excitation. Multi-echo imaging techniques extend this by acquiring images at various echo times (TEs) during the same excitation, sampling the T2* decay curve in more detail for each voxel and each repetition time (TR). Early functional spectroscopy experiments demonstrated contrast-to-noise ratio (CNR) benefits of multi-echo datasets [1]. Further investigation of multi-echo fMRI strategies has led to development of various weighting schemes to combine echoes for optimal BOLD contrast sensitivity [2,3]. More recently, a prototype multi-echo, coarse voxel (MECV) pulse sequence has been developed [4] and demonstrated in real-time neurofeedback experiments using existing weighting techniques [5]. Here, a novel method is presented for weighting multi-echo fMRI data using principal component analysis (PCA), validated by MECV-fMRI of a hand motor task.