The present study showed that the sagittal stratum in humans and monkeys has a complex fiber structure, beyond the well-known large fiber bundles running in rostro-caudal direction. We could not only clearly show the bilaminar structure of the sagittal stratum, but were able to show, where fibers enter and leave the sagittal stratum to the surrounding white matter on their way from/to different parieto-occipital cortical regions. These patterns as revealed here by 3D-PLI strikingly resembled those found in tracer studies in monkeys.
Escaping and joining fibers and layering of the sagittal stratum
A bilaminar configuration was found, reflecting two separate rostro-caudal fiber bundles, namely, the optic radiation in the external and the cortico-subcortical projection tracts in the internal layer (Flechsig
1896; Pfeifer
1925; Sachs
1892; Schmahmann and Pandya
2006). Supplementing these studies, the current data showed a clear layering into an external and an internal segment throughout the whole extent of the sagittal stratum. As has been described previously, the layering of the sagittal stratum dissolves in caudal sections close to the occipital pole, as does the sagittal stratum as a whole (Vergani et al.
2014). Approaching the occipital pole region means that less cortical regions were present, until only primary visual area V1 is visible. This phenomenon was particularly pronounced in the vervet monkey brain as V1 covers major parts of mesial and lateral aspects of the occipital lobe. In the human brain, V1 (Brodmann area 17) is mainly confined to the mesial side and only partially encroaches on surrounding occipital lobe at the very caudal end (Amunts et al.
2000). This suggests that only the optic radiation and the external layer of the sagittal stratum are present when approaching the occipital pole. The optic radiation then fans out to reach all parts of V1 and adjacent areas V2 and V3v and V3d (Alvaréz et al.
2015; Wandell and Winawer
2011), resulting in a disappearance of the sagittal stratum, since the fibers are not as parallel as in rostral sections.
The internal layer of the sagittal stratum appears in rostral sections as more and more fibers from occipital, parietal and temporal areas enter the sagittal stratum. These fibers necessarily need to cross through the external layer of the sagittal stratum. This has been clearly shown in tracer experiments in monkeys, where the fibers from a particular brain region can be followed through the white matter as they enter and cross the external layer of the sagittal stratum and join the fiber bundle running within the internal layer on their way to the subcortical target zones (Schmahmann and Pandya
2006). These tracing results have been used as reference in the present study. Shown for different exemplary tracts, the pattern of fiber architecture as revealed by the tracing experiments could be clearly shown by means of 3D-PLI. Of note, not only the global connection could be demonstrated, e.g., that fibers from superior and inferior parietal or ventral occipital cortex enter the sagittal stratum at different dorsal or ventral levels. In addition, the exact distribution of fibers within the sagittal stratum could be reconstructed. This was particularly visible for the fibers originating in the region of area V4 and V4D as compared to those coming from parietal areas. According to tracer experiments (Schmahmann and Pandya
2006), the fibers from V4/V4D enter the sagittal stratum in its middle sector in several parallel interspaced fiber bundles with latero-medial orientation, while the fibers from parietal areas entered the sagittal stratum in its dorsal sector as a large bundle with dorso-ventral orientation. This differential pattern of several interspaced vs. one large fiber bundle could be clearly shown in the 3D-PLI sections of the present study. Thus, without using a tracing technique with injections, 3D-PLI is able to show the exact same connection pattern in the same non-human primate lineage. Providing the same results as the gold standard for studying structural connectivity, e.g., of interspaced separated fiber bundles in the middle of the SSt, which was also reported in the tracer experiments of Schmahmann and Pandya (
2006), could be regarded as an external validation of the novel 3D-PLI technique. On the one hand, this facilitates systematic assessments of the complex white matter architecture and the different fiber bundles in one and the same brain (as compared to separate tracer injections in different hemispheres). On the other hand, it enables translating this approach to the human brain, where tracing using injections is ethically not feasible. 3D-PLI results obtained in the human brain could then be more easily understood as adequately revealing the fiber microarchitecture, which is a prerequisite for consecutive comparison and cross-validation with DWI data.
Thus, 3D-PLI could provide the bridge between tracing and diffusion imaging both within and across species (where comparable white-matter occipital fiber architecture can be detected, see, e.g., Kaneko et al.
2020), as the relevant intermediate step (Caspers and Axer
2019) and a valuable complement to comparisons between DWI and tracing experiments which also revealed comparable results (Schmahmann et al.
2007). At this, ex-vivo diffusion MRI serves as an additional relevant component when bridging between the scales (Roebroeck et al.
2019), as it could reach high resolutions on the mesoscale in the range of tens to hundreds of microns in different species (Aggarwal et al.
2015; Bech et al.
2020; Fritz et al.
2019; Liu et al.
2020; Ly et al.
2020; Wang et al. 2020).
It was also visible in the 3D-PLI data that these entering fibers indeed joined the internal layer of the sagittal stratum: while they could be clearly seen within the external layer, they faded out at the transition to the internal layer. This was indicative of fibers turning out of the sectioning plane into a perpendicular course, which in the case of our coronal brain sections pointed at a rostro-caudal trajectory. Following them continuously to their final subcortical target regions is challenging at the moment as fibers running out of the sectioning plane appear black in the 3D-PLI fiber orientation maps. This would be alleviated by 3D tractography in reconstructed stacks of 3D-PLI sections in the future.
We could additionally show that fibers crossed through both the external and internal segment of the sagittal stratum, originating in the medially adjacent tapetum. The tapetum contains the callosal fibers of the occipital and neighbouring parietal and temporal cortex. To reach their target zones within the cortex (Caspers et al.
2015; Clarke and Miklossy
1990), they have to cross through the sagittal stratum. We could show that they could be clearly distinguished from the fibers entering the sagittal stratum from the surrounding white matter, not only because of their major direction, but also by their complete course from leaving the tapetum, intermingling with and crossing through the different fibers of the sagittal stratum to leaving the sagittal stratum laterally.
Results obtained in this study rely on the analysis of one hemisphere of a vervet monkey brain and an exemplary comparison to a human brain section. Thus, future studies with a larger sample size are desirable to further verify the observations on the fiber architecture obtained here.
Resolving small fiber bundles in human occipital deep white matter
The packing density of fibers within the deep white matter of the human brain is even higher than in the monkey brain. This challenges current in-vivo diffusion imaging approaches, aiming at studying smaller bundles which cross the large fiber tracts (Dell’Acqua and Catani
2012; Jbabdi and Johansen-Berg
2011). This is particularly true within the occipital lobe, where several large fiber tracts run in parallel, mainly in rostro-caudal direction, such as the optic radiation (as part of the sagittal stratum), the inferior longitudinal fasciculus, the callosal fibers within the tapetum, or the inferior and superior fronto-occipital fasciculi (Catani et al.
2003; Cristina et al.
2014; Forkel et al.
2014; Takemura et al.
2020). Using additional MR contrasts, it was possible to identify another relatively large fiber tract perpendicular to them, running in dorso-ventral direction lateral to the sagittal stratum and crossing through the ILF, i.e., the vertical occipital fasciculus (Bugain et al.
2021; Takemura et al.
2016; Yeatman et al.
2014). These large bundles surrounding the sagittal stratum have also been identified in blunt dissections (De Benedictis et al.
2014).
When it comes to the fine structure beyond these large tracts, e.g., within the sagittal stratum, both dissection and diffusion imaging seemed to have reached their limits in terms of resolution: the bilaminar structure of an internal and external sagittal stratum could only partially be identified (Vergani et al.
2014).
Functional in-vivo mappings during neurosurgery suggest that several complex higher order functions (e.g., hemi-agnosia, spatial neglect, known to involve several occipital and parietal cortical regions) were hampered during direct electrical stimulation of the SSt, indicating involvement of different fiber systems intermingling within the SSt (Berro et al.
2021). The current results shed new light on this issue: the fibers enter, leave or cross through the sagittal stratum in very thin bundles or even as single fibers. This is far below the spatial and angular resolution of routine in-vivo diffusion imaging studies, even of high-angular resolution diffusion protocols, when it comes to identifying single fibers or disentangling crossings of such small bundles. Using advanced fiber orientation reconstruction algorithms, such as constrained spherical deconvolution (Tournier et al.
2007), might enable detection of some additional fiber orientations, but averaged across a considerably large volume of a few cubic millimeters.
The analysis of ultra-high-resolution images resulting from 3D-PLI showed that there are indeed small fiber bundles which cross through the inferior longitudinal or vertical occipital fasciculus on their way to or from the sagittal stratum. They could not only clearly be followed as they crossed through the external sagittal stratum to join the internal sagittal stratum, but also as they approached the sagittal stratum and crossed the large white matter tracts surrounding it.
Thus, the additional fiber directions as visible in the diffusion imaging data (Fig.
1C,
D) might indeed hint at such anatomical fine structure of the sagittal stratum. Considering the reconstructed fiber orientations from the diffusion signal alone would prevent any distinction between a correct depiction of the underlying anatomy or false positives due to noise in the data (see, e.g., Leuze et al.
2021). Combination with an ultra-high-resolution technique such as 3D-PLI may provide detailed insights into the fiber architecture of the brain by decoding the local fiber configurations (Caspers and Axer
2019). This particularly holds true for regions with one major fiber direction dominating the local architecture, such as the sagittal stratum, which could lead to neglect of the minor directions. Based on the examples shown here for the human brain with a highly similar pattern of fiber architecture as found in the vervet monkey, future studies are needed to follow the individual fiber tracts which join the internal layer of the sagittal stratum to reach their subcortical target regions. This might be facilitated in the future by fiber tractography based on reconstructed stacks of 3D-PLI sections, based on aggregated 3D-PLI-derived fiber orientation distribution functions (Alimi et al.
2020; Axer et al.
2016; Reckfort et al.
2015; Schmitz et al.
2018), in combination with ultra-high-field ex-vivo DWI (Aggarwal et al.
2015; Fritz et al.
2019; Ly et al.
2020), as has been shown, e.g., for the trigeminal pathway in the human brainstem (Henssen et al.
2019). With ex-vivo DWI being inherently 3D when applied to a whole tissue volume, it could serve as relevant reference for the tractography data derived from 3D-PLI. With the high resolution of 3D-PLI of about one micrometer in-plane across species, these approaches are mutually complementary and could thus be used in combination in the future to understand the complex multi-level fiber architecture of the brain.
Similar to the knowledge gained from respective tracing experiments in monkeys, this high-resolution fiber tract analysis in humans is of particular importance to understand which parts of occipital and adjacent parietal and temporal cortices are connected with which subcortical structure, e.g., within the thalamus or among the pontine or mesencephalic nuclei. This will enable a better understanding of cortico-subcortical feedback loops for top-down control of incoming information from the subcortical relay stations as well as information flow to subcortical control centers.