Deconstructing white matter connectivity of human amygdala nuclei with thalamus and cortex subdivisions in vivo

Abstract

Structural alterations in long-range amygdala connections are proposed to crucially underlie several neuropsychiatric disorders. While progress has been made in elucidating the function of these connections, our understanding of their structure in humans remains sparse and non-systematic. Harnessing diffusion-weighted imaging and probabilistic tractography in humans, we investigate connections between two main amygdala nucleus groups, thalamic nuclei, and cortex. We first parcellated amygdala into deep (basolateral) and superficial (centrocortical) nucleus groups, and thalamus into six subregions, using previously established protocols based on connectivity. Cortex was parcellated based on T1-weighted images. We found substantial amygdala connections to thalamus, with different patterns for the two amygdala nuclei. Crucially, we describe direct subcortical connections between amygdala and paraventricular thalamus. Different from rodents but similar to non-human primates, these are more pronounced for basolateral than centrocortical amygdala. Substantial white-matter connectivity between amygdala and visual pulvinar is also more pronounced for basolateral amygdala. Furthermore, we establish detailed connectivity profiles for basolateral and centrocortical amygdala to cortical regions. These exhibit cascadic connections with sensory cortices as suggested previously based on tracer methods in non-human animals. We propose that the quantitative connectivity profiles provided here may guide future work on normal and pathological function of human amygdala. Hum Brain Mapp 38:3927-3940, 2017. © 2017 Wiley Periodicals, Inc.

Keywords: DWI; connectome; neuroanatomy; paraventricular; probabilistic tractography; structural connectivity.

Figure 1

Coronal, axial and sagittal views of group‐level amygdala parcellations in MNI space. Maximum probability maps in upper panels are thresholded at P = 0.3 (superficial cluster red, deep cluster blue). Unthresholded probability maps for basolateral and centrocortical clusters in lower panels.

Figure 2

A: Cortical masks for thalamus segmentation, based on FreeSurfer cortical parcellation (upper left panel, example subject). The 3D rendered maximum probability maps of corresponding group‐level thalamus segmentations in MNI space (thresholded at P = 0.3). B: Unthresholded probability maps of group‐level thalamus segmentations in MNI space. C: Axial and coronal views of group‐level paraventricular mask in MNI space on the left. 3D visualization of group‐level PVT mask (purple) and temporal thalamic segmentation (thresholded at P = 0.3).

Figure 3

Mean [± SEM] connection strength from basolateral and centrocortical amygdala cluster to thalamic parcellations. ANOVA in Table I. Post‐hoc t‐tests: ** P < 0.01; *** P < 0.001.

Figure 4

Mean [± SEM] connection density from basolateral and centrocortical amygdala cluster to thalamic parcellations (i.e., connection strength corrected for target size). ANOVA in Table I. Post‐hoc t‐tests * P < 0.05; ** P < 0.01; *** P < 0.001.

Figure 5

Voxel‐by‐voxel analysis of mean connection strength between A: basolateral or B: centrocortical amygdala cluster and cortical surface. Color‐coding is on logarithmic scale.

Figure 6

Mean [± SEM] connection strength from basolateral and centrocortical amygdala cluster to cortex. The 12 cortical areas with the highest connection strength to the amygdala are shown. Cortex areas used in amygdala parcellation (i.e., OFC/TP) are marked red. ANOVA in Table I. Post‐hoc t‐tests, * P < 0.05; ** P < 0.01; *** P < 0.001. [Color figure can be viewed at http://wileyonlinelibrary.com]

Figure 7

Mean [± SEM] connection density from basolateral and centrocortical amygdala cluster to cortex (i.e. connection strength corrected for target size). The 12 cortical areas with the highest connectivity to the amygdala are shown. Cortex areas used in amygdala parcellation (i.e., OFC/TP) are marked red. ANOVA in Table 1. Post‐hoc t‐tests, * P < 0.05; ** P < 0.01; *** P < 0.001. [Color figure can be viewed at http://wileyonlinelibrary.com]