This study demonstrates that perfusion fixation substantially alters whole-brain structural connectivity in a region-specific manner. Nearly 15% of all inter-regional connections showed significant changes following fixation, with connectivity alterations closely associated with regional volume changes. These findings advance our understanding of fixation effects from local microstructural changes to macroscopic network-level alterations, providing critical reference data for interpreting ex vivo tractography studies.

The observed connectivity alterations likely result from multiple interacting mechanisms. Perfusion fixation causes non-uniform tissue contraction [7, 8], which alters geometric relationships between brain regions. This mechanical deformation may change the tractography-derived length and curvature of white matter pathways, potentially affecting fiber orientation distribution estimation and subsequent tractography reconstruction. For instance, elongated or more tortuous pathways could accumulate greater tractography uncertainties during streamline propagation. The observation that regions with greater volume changes exhibited more connectivity alterations suggests that tissue shrinkage is closely associated with connectivity changes, although the precise mechanisms underlying this relationship remain to be elucidated. In addition to geometric changes, altered diffusion properties likely contribute to connectivity changes. Fixation modifies tissue microstructure, changing water diffusion characteristics that form the basis of tractography algorithms [9, 10]. Using diffusion tensor imaging, we previously demonstrated that axial diffusivity, a diffusion tensor metric reflecting water diffusion along the principal fiber direction, is particularly sensitive to regional volume changes following fixation [13]. Although the current study employed CSD-based analysis rather than tensor-based metrics, fixation-induced changes in water diffusion properties would similarly affect fiber orientation distribution estimation. These local diffusion alterations propagate through the tractography pipeline, ultimately affecting connectivity estimates. Both geometric and diffusion-related factors may interact in complex ways, and the relative contribution of each mechanism warrants further investigation.

Beyond these macroscopic geometric and diffusion-related factors, recent work by Santini et al. investigated microstructural changes between in vivo and perfusion-fixed ex vivo marmoset brains using advanced diffusion MRI techniques including oscillating gradient and b-tensor encoding at 9.4T [27]. Their findings demonstrated that perfusion fixation induces substantial microstructural alterations, including decreased extracellular volume fraction, potential axon beading, and increased dot compartment signal fraction. These microstructural changes at the cellular level would further affect fiber orientation distribution estimation and tractography outcomes, complementing the macroscopic tissue deformation discussed above. It should also be noted that the perfusion fixation process itself involves a brief period of blood washout before fixative reaches the tissue, during which transient ischemic conditions may occur. Temporal and frontal regions are known to be particularly vulnerable to ischemic insult due to their vascular anatomy. Although transcardial perfusion in our protocol was performed immediately following deep anesthesia to minimize such effects, we cannot entirely exclude the possibility that perfusion-related tissue changes contributed to the regional patterns of connectivity alteration observed in this study.

Temporal and limbic regions, including the temporopolar area and hippocampal formation, showed the greatest connectivity changes. These areas contain complex fiber pathways with varying curvature, and such complex fiber architecture may interact with fixation-induced tissue changes to produce greater tractography variability. In contrast, regions with minimal volume changes, such as the piriform cortex and superior colliculus, showed fewer connectivity alterations. This pattern suggests that the extent of local tissue deformation is closely associated with the degree of fixation-induced connectivity changes.

The preferential effect on connections between anatomically distant regions, particularly those involving temporal and frontal areas, further supports the role of geometric factors. Longer pathways traversing multiple brain regions may accumulate greater tractography uncertainties when tissue geometry is distorted.

The observation that connections between neighboring regions showed increased connectivity following fixation may reflect differential effects of tissue shrinkage on local versus distant pathways. Tissue contraction could increase local white matter density, potentially affecting fiber orientation distribution estimation and SIFT2 weighting in ways that differ from longer pathways. However, the precise mechanisms underlying this phenomenon remain unclear and warrant further investigation.

These findings have important implications for the growing field of ex vivo connectomics [5, 6]. First, direct comparison between in vivo and ex vivo tractography should account for systematic connectivity biases, particularly in temporal and limbic networks. Second, cross-species comparisons using ex vivo data may be confounded by differential fixation effects across brain regions. Third, validation of in vivo tractography using ex vivo data requires careful consideration of which connections are most affected by fixation.

The observation that approximately 15% of connections are significantly altered represents a substantial effect that cannot be ignored in ex vivo studies. Researchers using fixed tissue for tractography should consider these biases when interpreting their results.

It should be noted that the diffusion MRI acquisition parameters differed between in vivo and ex vivo conditions in the present study. These differences were intentional to optimize data quality for each condition, as ex vivo tissue exhibits substantially reduced diffusion coefficients requiring higher b-values for adequate signal detection [19, 20, 28]. Importantly, the ex vivo acquisition employed parameters that are generally more favorable for tractography, including higher angular resolution and finer spatial resolution. Despite these more favorable acquisition conditions, the majority of significantly altered connections showed decreased rather than increased connectivity in the ex vivo condition. This directional pattern suggests that the observed connectivity changes reflect genuine biological effects of fixation rather than artifacts arising from differences in acquisition parameters.

Several limitations should be acknowledged. First, we examined only one fixation protocol using 4% paraformaldehyde, and different fixatives or concentrations might produce different patterns of change. Second, our findings in marmosets may not directly translate to species with different brain sizes or white matter organization. Third, we did not examine the temporal dynamics of connectivity changes during the fixation process. Fourth, while our spatial resolution was high for in vivo standards, subtle connectivity changes might not be detectable at the voxel sizes employed. Fifth, we did not directly quantify geometric changes in white matter pathways such as length and curvature. The extent of fixation-induced geometric deformation likely depends on multiple factors including fixation protocols and species-specific tissue properties. Sixth, in the present study, all tractography analyses were performed in each animal’s native diffusion space, with the atlas parcellation transformed into this space for connectivity matrix construction. An alternative approach would be to register both in vivo and ex vivo data to a common template space to potentially isolate microstructural diffusion changes from macroscopic geometric deformation. However, spatial normalization of diffusion-weighted data introduces interpolation artifacts that contaminate voxel-level diffusion indices through merging and averaging of neighboring voxel information, and this effect becomes more pronounced with larger deformations such as those induced by fixation-related tissue shrinkage. For this reason, performing tractography in native space is the standard approach in diffusion MRI connectivity analysis. Separating the relative contributions of macroscopic geometric changes and microscopic diffusion property alterations remains an important challenge for future studies, potentially through the development of methods that do not require spatial transformation of diffusion data [27, 28].