Developing a diceCT brain atlas: Comparison of diceCT neuroanatomy and brain histology in a rat model

Abstract

Significant advancements and refinement of neurological imaging modalities can enable new understanding in nervous system organization. One recent example, diffusible iodine‐based contrast‐enhanced computed tomography (diceCT), is an iodine staining and X‐ray μCT‐imaging technique that allows for the differentiation of myelinated from unmyelinated nervous tissues at finer scales of spatial resolution than standard three‐dimensional (3D) brain‐imaging tools allow. DiceCT is versatile, capable of imaging specimens across several orders of magnitude in size. It enables large and complex structures such as human brains to be studied alongside those of less complex nervous systems (e.g., fish, frogs, rodents, birds). One key feature is that the staining agent can be removed, allowing the subsequent application of acetate, fluorescent, and immunofluorescent histology following diceCT imaging. Nevertheless, the relative novelty of diceCT as a neuroanatomical visualization tool also demands ground‐truthing by determining its practical reliably for distinguishing precise boundaries between brain structures. For example, some adjacent, tissue‐level neuroanatomical features are poorly differentiated due to staining similarity or inherently obscure boundaries (such as the thalamus, which houses many small white‐matter tracts). To address this, we compare 3D orientation‐matched image stacks of diceCT juvenile Sprague Dawley rat brains to the Paxinos and Watson atlas, a gold‐standard rat‐brain reference. Visual identification of brain structures was attempted for all brain regions. Cardinal boundaries for each structure's dorsal, ventral, medial, and lateral edges were scored as distinguishable or non‐distinguishable. Right/left staining asymmetries were also scored. We used this outline to compare and contrast the neuroanatomical structures identified by diceCT with those described by Paxinos and Watson. Preliminary results indicate that 42% of all scored structures were visible via most (e.g., 3–4) of their cardinal boundaries, whereas 25% of atlas structures showed no distinguishable boundaries using diceCT. Overall, diceCT rat‐brain sections align well with standard histological sections, and visibility of neuroanatomical features is similar to tissue‐based, 3D magnetic resonance imaging atlases of comparable resolution.