Robert J Anderson^{1}, Christopher Long^{1}, Evan D Calabrese^{2}, Scott H Robertson^{1}, Gary P Cofer^{1}, G Allan Johnson^{1}, and Alexandra Badea^{1}

Network approaches provide sensitive biomarkers for neurological conditions such as Alzheimer’s disease. Mouse models provide tools to dissect vulnerable circuits at prodromal stages, and to assess the effects of interventions. We have simulated mouse brain structural connectomes, balancing angular, spatial resolution and scan time. Specifically, we evaluated protocols with 6, 12, 15, 20, 30, 45, 60 and 120 angles; and 3 voxel sizes at 43, 86 and 172 µm. Our results indicate schemes using 46 or 60 diffusion directions, acquired at 86 µm resolution achieve a good cost/performance balance relative to a high spatial, high angular resolution sampling scheme.

Figure 1. The effect
of spatial resolution (horizontal axis) and angular resolution (vertical axis)
on the number of voxels with 1, 2, 3, or 4 fibers (dyads) per voxels. (A)
illustrates these effects for 12, 45 and 120 directions, chosen as examples for
low, medium, and high angular sampling. (B) illustrates the effects for
sampling schemes between 12 and 120, at 3 spatial resolutions (43 µm, 86 µm,
and 172 µm). Changing resolution affected the first dyads, and was more
prominent for the higher (2nd) order dyads. These results illustrate
the advantages of high angular and spatial resolution protocols.

Figure 2. The
connectivity of the fimbria was reconstructed for protocols with 12, 45, 120
directions, at the full 43 µm resolution, illustrating limitations for the
small 12 directions protocol.

Figure 3.
Dyad 1 dispersion showed decreasing errors for higher angular samples, with an
inflection point at 45°, and increasing stability after 60° (A). Similar errors
were apparent for 43 and 86 µm, but larger for 172 µm. Dyad2 showed a higher
effect of resolution, and larger errors, tapering off after 60° (B).

Figure 4. The adjacency matrices for a subset of 13
regions relevant to neurodegenerative diseases showed greater similarity to the
reference connectome for 45 directions, compared to 12 directions (A). This can
be quantified through a graph similarity matrix (B). The chord diagrams (C) shows that all
protocols captured the connectivity of the hippocampus (HC) and its connecting
fibers (fx: fornix, fi: fimbria), but shorter protocols had reduced sensitivity
for smaller nuclei (LD: laterodorsal thalamic nuclei, LGN: lateral geniculate
nuclei; LPO: lateral preoptic nucleus), and lacked thesensitivity required for cortical-cortical
connectivity e.g. for the cingulate (Cg) and entorhinal cortices (Ent).