Spatial Resolution and Imaging Encoding fMRI Settings for Optimal Cortical and Subcortical Motor Somatotopy in the Human Brain

Abstract

There is much controversy about the optimal trade-off between blood-oxygen-level-dependent (BOLD) sensitivity and spatial precision in experiments on brain's topology properties using functional magnetic resonance imaging (fMRI). The sparse empirical evidence and regional specificity of these interactions pose a practical burden for the choice of imaging protocol parameters. Here, we test in a motor somatotopy experiment the impact of fMRI spatial resolution on differentiation between body part representations in cortex and subcortical structures. Motor somatotopy patterns were obtained in a block-design paradigm and visually cued movements of face, upper and lower limbs at 1.5, 2, and 3 mm spatial resolution. The degree of segregation of the body parts' spatial representations was estimated using a pattern component model. In cortical areas, we observed the same level of segregation between somatotopy maps across all three resolutions. In subcortical areas the degree of effective similarity between spatial representations was significantly impacted by the image resolution. The 1.5 mm 3D EPI and 3 mm 2D EPI protocols led to higher segregation between motor representations compared to the 2 mm 3D EPI protocol. This finding could not be attributed to differential BOLD sensitivity or delineation of functional areas alone and suggests a crucial role of the image encoding scheme - i.e., 2D vs. 3D EPI. Our study contributes to the field by providing empirical evidence about the impact of acquisition protocols for the delineation of somatotopic areas in cortical and sub-cortical brain regions.

Keywords: BOLD sensitivity; functional magnetic resonance imaging; image resolution; segregation; subcortical areas.

FIGURE 1

Motor somatotopy patterns across resolutions and brain regions projected on canonical anatomical image in standard space. Group results obtained using the flexible factorial design showing the binarised statistical parametric maps (t-values) thresholded at α = 0.05 (corrected for multiple comparisons, family-wise error rate) for each resolution. Left and right body movements are merged (red – feet; yellow – hands; green – face).

FIGURE 2

Z-scores of IoS for hand against foot (Ha. vs. Fo.), hand against face (Ha. vs. Fa.) and foot against face (Fo. vs. Fa.) per resolution and ROI, projected on canonical anatomical image in standard space. Bar plots on the left are for left ROIs, bar plots on the right are for right ROIs. Surface renderings of the putamen (magenta), pallidum (orange), motor nuclei of the thalamus (cyan), SMA (mid-tone blue) and M1 (violet). EPI protocols denoted by hatched light gray – 1.5 mm; mid-tone gray – 2 mm, black – 3 mm). Stars indicate significantly different Z-scores (p < 0.05, FDR-corrected).

FIGURE 3

Z-scores of IoS pair-wise comparison for 1.5 mm, 2 mm and 3 mm resolution for different body parts – hands, feet, face and regions-of-interest, projected on canonical anatomical image in standard space. Regions-of-interest in putamen (magenta), pallidum (orange), motor nuclei of the thalamus (cyan), SMA (mid-tone blue) and M1 (violet). Dotted lines indicate significance of correlation (p < 0.05 uncorrected for multiple comparisons, bilateral test).

FIGURE 4

IoS values per ROI as a function of tSNR. Light gray line: regression line across all image resolutions, significance denoted with red stars; thick black line: regression line for 3 mm data; thick dark gray: regression line for 2 mm data, significance denoted with green stars; dotted line: regression line for 1.5 mm data.