When tractography meets tracer injections: a systematic study of trends and variation sources of diffusion-based connectivity

Abstract

Tractography is a powerful technique capable of non-invasively reconstructing the structural connections in the brain using diffusion MRI images, but the validation of tractograms is challenging due to lack of ground truth. Owing to recent developments in mapping the mouse brain connectome, high-resolution tracer injection-based axonal projection maps have been created and quickly adopted for the validation of tractography. Previous studies using tracer injections mainly focused on investigating the match in projections and optimal tractography protocols. Being a complicated technique, however, tractography relies on multiple stages of operations and parameters. These factors introduce large variabilities in tractograms, hindering the optimization of protocols and making the interpretation of results difficult. Based on this observation, in contrast to previous studies, in this work we focused on quantifying and ranking the amount of performance variation introduced by these factors. For this purpose, we performed over a million tractography experiments and studied the variability across different subjects, injections, anatomical constraints and tractography parameters. By using N-way ANOVA analysis, we show that all tractography parameters are significant and importantly performance variations with respect to the differences in subjects are comparable to the variations due to tractography parameters, which strongly underlines the importance of fully documenting the tractography protocols in scientific experiments. We also quantitatively show that inclusion of anatomical constraints is the most significant factor for improving tractography performance. Although this critical factor helps reduce false positives, our analysis indicates that anatomy-informed tractography still fails to capture a large portion of axonal projections.

Whole mouse brain connectomics

“…Chen et al () compared DTI data in a mouse brain acquired at 62.5 μm (0.0625 mm 3 ) with six angles to the ABCA with RI = 24,576. A more recent study from Aydogan et al () acquired 90 angles at 200 μm resolution, yielding an RI of 11,250. Azadbakht et al () compared dMRI to previously published schemata for interareal connectivity based on neuro histological tract‐tracing methods in the macaque visual cortex; in their study, RI was 1,509 (120 angles and 0.43 mm spatial resolution).…”

Toward an MRI-Based Mesoscale Connectome of the Squid Brain

“…It is worth underlining that, although MRI-based tractography has previously attracted a great deal of criticism (Thomas et al, 2014;Maier-Hein et al, 2017;Aydogan et al, 2018;Sotiropoulos and Zalesky, 2019), the methods and cross-checks applied here enable us to suggest that all new tracts proposed are valid (and see Calabrese et al [2015]). Specifically, in a comparison with the previously known 282 squid neural tracts described by Young and his colleague , 281 were refound using the high angular resolution diffusion-weighted imaging (HARDI) method combined with an ultra-conservative level for tractography acceptance employed here.…”

“…ideas are applied here to map and model the elements and potential interactions of the squid brain (Bullmore and Sporns, 2009;Bassett and Sporns, 2017;Assaf et al, 2019). As noted in mouse and other vertebrate work (Jbabdi and Johansen-Berg, 2011;Donahue et al, 2016;Maier-Hein et al, 2017;Aydogan et al, 2018;Suarez et al, 2018;Sotiropoulos and Zalesky, 2019), several caveats should be applied to probabilistic MRI-based tractography, and it is often best viewed as supportive and predictive rather than a fully validated functional connectome.…”

Optimizing Diffusion Imaging Protocols for Structural Connectomics in Mouse Models of Neurological Conditions

“…While studies on diffusion imaging in the human brain have previously addressed the tradeoffs associated with spatial vs. angular resolution, the mouse brain provides insight into white matter connectivity at a vastly different scale and, thus, deserves special attention. We need to compare 2-mm linear dimension voxel sizes in humans (corresponding to 8-mm 3 voxel volumes), vs. 0.2-0.043-mm linear dimension voxel sizes (corresponding to ∼8 × 10 −2 -8 × 10 −5 -mm 3 voxel volumes) required to distinguish similar levels of anatomical detail in mouse brains [6][7][8][9][10][11]. This increased resolution enables the observation of brain architecture at the level of cellular layers, which, therefore, can help us gain insight into the mechanistic drivers behind the etiology and progression of disease, using animal models where axonal dimensions are relatively similar to humans [12,13].…”