Simultaneous functional MRI of two awake marmosets

Abstract

Social cognition is a dynamic process that requires the perception and integration of a complex set of idiosyncratic features between interacting conspecifics. Here we present a method for simultaneously measuring the whole-brain activation of two socially interacting marmoset monkeys using functional magnetic resonance imaging. MRI hardware (a radiofrequency coil and peripheral devices) and image-processing pipelines were developed to assess brain responses to socialization, both on an intra-brain and inter-brain level. Notably, the brain activation of a marmoset when viewing a second marmoset in-person versus when viewing a pre-recorded video of the same marmoset-i.e., when either capable or incapable of socially interacting with a visible conspecific-demonstrates increased activation in the face-patch network. This method enables a wide range of possibilities for potentially studying social function and dysfunction in a non-human primate model.

Fig. 1. Mechanical setup of the social coil.

Each marmoset is placed in a restraint device with an integrated radiofrequency coil. Initially, the arms of the restraint device are fully opened to allow the marmoset to enter a tube and be restrained by lockable neck and tail plates. Coil elements are located on the inner surface of two hinged arms; circuit boards interfacing to the coil elements are adhered to the top of the hinged arms. Once the marmosets’ bodies are restrained, they are placed on a custom platform that allows reproducible and variant positioning within the scanner. Gel is subsequently placed on the marmosets’ heads to reduce susceptibility artifacts and geometric distortion. Marmosets are then head-fixed by closing the hinged coil arms, inserting the locking pins, and fully extending the coil clamp with a tightening screw. This creates a four-point fixation of the chamber and electrically completes the coil element circumscribing the chamber by pressing conductive screws into opposing conductive pads. Preamplifier modules are repositioned independent of the coil to ensure low-noise amplifiers always maintain their optimal orientation with respect to the main magnetic field regardless of coil rotation. In this study, marmosets were placed 11-cm apart and facing each other: a distance chosen to allow natural social interaction based on the visual acuity of the marmoset. A camera was employed to monitor marmosets during scanning. C_T: tuning capacitor; C_M: matching capacitor; RFC: radiofrequency choke; DC: direct current.

Fig. 2. Signal and noise characteristics of the social coil.

a Intra- and inter-coil noise correlation of the two receive arrays corresponding to the coil-element layout and numbering depicted in the upper inset (represented in a planar view: coil 1: elements 1–5; coil 2: elements 6–10). The lower inset refers to inter-coil noise correlation, which has a maximum of 2.3%. b Temporal SNR of marmoset M1 (coil 1) and marmoset M3 (coil 2) when face-to-face along the z-axis of the scanner. ROIs (dashed regions) in the sagittal, axial, and coronal planes show inter-brain regional differences of less than 10%, which is less than intra-brain differences (which are approximately 2 to 3-fold). Maps have been reoriented into radiological convention to facilitate comparison. c Histograms of the temporal SNR of each voxel within the brain of each marmoset show similar distributions. Dashed lines represent the mean temporal SNR for each marmoset/coil combination.

Fig. 3. Image distortion due to local magnetic field inhomogeneities.

The unique dual-marmoset setup has implications on image distortion during fMRI acquisitions: face-to-face marmosets experience opposing phase-encode directions, which creates different geometric distortion. To compensate, echo-planar images are acquired with alternating phase-encode directions. In this topology, marmoset M1, with left-right (LR) phase encode, will have similar distortion to marmoset M2 with right-left (RL) phase-encode. The most notable image distortion is found in the temporal poles and at the boundary of the chamber. Successive functional runs are then used to estimate the off-resonance correction field required for correcting image distortion. After functional images are distortion-correction and brain-extracted, they are registered to an anatomical image (i.e., anatomical space). Anatomical images, having been registered to the NIH marmoset brain atlas, are used to register the functional images (in anatomical space) onto the brain atlas (template space). Registration of functional images to the marmoset brain atlas facilitates the assignment of localized brain activation to known networks. Grey- and white-matter boundaries are shown as blue and yellow lines, respectively.

Fig. 4. Intra-brain activation maps derived from the simultaneous scanning of two marmosets within each other’s visual field.

The social coil provided sufficient temporal SNR and resolution to discern seven functional networks in marmosets M1 and M3. Three representative functional networks are displayed: the ventral somatomotor network (SMNv), dorsal SMN (SMNd), and medial SMN (SMNm). Networks are displayed as z-score maps on the cortical surface, with only the left hemisphere visible. White lines represent cytoarchitectonic borders. Supplementary Fig. 1 shows all remaining networks in both hemispheres and at the volume level.

Fig. 5. Synchronous brain activation between two marmosets within each other’s visual field.

Two marmosets, M1 and M3, were placed facing each other while acquiring four functional time courses. The correlation coefficient between the time courses of spatially analogous voxels in marmosets M1 and M3 was calculated and transformed to a z-score map—z-scores are presented on a flattened map of the left and right hemispheres, with black lines indicating cytoarchitectonic borders. Synchronous activity was found in the anterior cingulate cortex, ventromedial prefrontal cortex, and temporal cortex—regions thought to play an important role in social interaction.

Fig. 6. Preferentially activated brain regions in two marmosets within each other’s visual field.

a Two task-based fMRI paradigms were acquired. In paradigm 1, a marmoset (M4) could view a second marmoset (M3) in-person—i.e., it was capable of socially interacting with a visible conspecific. In paradigm 2, M4 viewed a pre-recorded video of M3—i.e., it was incapable of socially interacting with a visible conspecific. In each paradigm, the two marmosets (M3 being in-person or a video) were placed face-to-face, 11-cm apart, and separated by smart films. The opacity of the smart films was alternated in a block design to permit or deny visual contact. b Regions of increased brain activation were derived for each paradigm by comparing the two epochs of the block design in a using a two-sided t-test. Activation maps are represented on the left and right fiducial surface of the M4 marmoset brain (t-scores >2.26, p < 0.05, uncorrected for multiple comparisons), with white lines representing cytoarchitectonic borders. c The equivalent activation maps as represented on coronal slices to highlight subcortical activations. d A voxel-wise, two-sided t-test between the stimulus conditions of paradigms 1 and 2 indicates regions of increased activation due to socialization (z-scores > 1.96, p < 0.05, uncorrected for multiple comparisons). The displayed sagittal slice indicates the position of coronal slices.