Since the early 2000s, the expansion of digital anatomy tools has been aided by advances in computational power and accessibility of medical imaging such as Computed Tomography (CT). The greater accessibility to digital imaging of fossil material has allowed the reconstructions of inner cranial cavities (endocasts), sinuses cavities, and dental reconstructions of the enamel-dentine junctions (EDJ) of fossilized teeth. Despite great accessibility, the segmentation processes used to generate digital reconstructions of inner cavities remain time-consuming and require specific expertise in computer analysis, anatomy, digital imaging.
Profico et al. (2018) provide two fully-automated digital methods to minimize these time-consuming digital segmentation tasks. Both of these methods rely on point-of-views (POVs) to delineate a region-of-interest (ROI). In the CA-LSE method, POVs were located outside a ROI and all areas beyond are subtracted from the final reconstruction. In contrast, the AST-3D method relies on a ROI defined by POVs placed inside a cavity and all external areas, subtracted from the final reconstruction. While both methods are similar and can be used to generate reconstructions of the inner cavities, each method has slightly different benefits. Profico and colleagues conducted a comparison of both methods to determine strengths and weaknesses of each approach. While both of these methods are available through the Cran R network, two different R packages were tested: Morpho and Arothron.
Results indicated that in the Morpho package, CA-LSE had no restrictions on where POVs could be placed, but using AST-3D method in Morpho, POVs had to be manually placed inside the internal cavity for successful reconstruction. In the Arothron package, CA-LSE method allowed fully-automated placement of POVs outside the ROI surface, however, the AST-3D method a ROI must be defined by manually placed POVs within the inner cavity. In general, accuracy of the AST-3D and CA-LSE methods were determined by each method, with AST-3D more reliable generating reconstructions of inner cavities (such as endocasts), while the CA-LSE was more suited to reconstructions of outer structures (such as skulls).
Although, automatic approaches offer time-efficiency and often allow larger sample sizes to be more quickly processed, many fossilized skulls are highly fragmentary and automated methods remain limited when fossilized remains are partially or entirely matrix-filled with anatomical and digital expertise still requiring manual segmentation. In these complex scenarios, further fine-tuning of automated methods would be invaluable with inclusion of fully-automated, semi-automatic and manual options.
European Society for the study of Human Evolution – Faro, 2018
Duan and colleagues developed a new computational method for automatic detection of patterns of cortical folding in large datasets. This method extracts multiple features that characterize the folding patterns, such as sulcal bottoms and gyral crest curvatures. Then, an overall similarity matrix is calculated that contains information on shared patterns and individual variation. Finally, the subjects are clustered on groups that represent a common folding pattern. The authors show their method is more efficient than previous ones in detecting folding patterns and clustering subjects into affinity groups. They validate its reproducibility and reliability in two main samples. They demonstrate the application of their methodology on a large sample of 595 healthy neonate brains, to characterize folding patterns in newborns. Then, they compare their results to a dataset of adult brains from the Human Connectome Project. They focus on four cortical regions, the superior temporal gyrus (STG), the inferior frontal gyrus (IFG), the cingulate cortex, and the precuneus, considering both sex differences and hemispheric asymmetries. Overall, the typical folding patterns of infants were consistent with those of adults, evidencing that cortical folds are largely established from an early age. On the other hand, some differences were also identified. For instance, four folding patterns were recognized in infant STG, while adults have an extra pattern. In contrast, one of the neonates’ IFG pattern is absent in adults. In both samples, there are sex differences in the proportions of some of the folding patterns of STG, IFG, and cingulate cortex. Hemispheric asymmetries were observed in the cingulate and STG, being more significant in the latter, which the authors suggest might reflect language lateralization. Considering the precuneus, their method revealed three main gyral patterns which were not associated with sex differences or hemispheric asymmetries. These gyral patterns appear to be mainly grouped based on the presence of either a gyral structure (patterns 1 and 2, more frequent) or a deep sulcus (pattern 3) in the middle of the precuneus. According to the authors, these groups are similar to the ones described in a previous study by our lab on a sample of adult healthy brains.
From Fossils to Function
Integrative and Taxonomically-Inclusive Approaches to Vertebrate Evolutionary Neuroscience
Brain, Behavior and Evolution, 91
Comparative neurobiology has traditionally been used to describe and quantify the macro-and micro-anatomical changes to the brain between human and non-human primates. Research literature commonly refers to the Stephen dataset with brain volumes measured from a small sample of ex-vivo non-human primate brains with species often represented by only one or two individuals. Although this is common limitation inherent to many physical neuroanatomy collections, caution should be used with such limited samples not representative of a species and the actual variation unknown. The consequences for such assumptions on quantifying the neuroanatomical differences between humans and non-human primates have broader implications for human evolution. Despite the increasing accessibility of primate neuroimaging datasets, many comparative studies still rely on the Stephen brain volumes. This is despite the necessary factoring of numerous bias including cerebral tissue damage from the delay between the post-mortem interval and brain preservation, potential introduction of artifacts from tissue preservation processes causing shrinkage, some cellular destruction and occasional damage during brain extraction.
Navarette and colleagues recently compared digital neuroanatomical volumes from ex-vivo brain MRI with the Stephen data using the same primate species but including an extra 20 species. Results showed differences between the Stephen data and those obtained by Navarrete et al. with larger brain volumes measured in the pre-fixed state versus post-fixed, indicating fixation did noticeably affect brain volume measurement. Although Navarrete et al. aimed to quantify primate brain volume variation by increasing the number of primate species to 39, there were still 29 species represented by only one individual, while the maximum of for the entire sample was never greater than three individuals per species. Although Navarrete et al. argued a lack of larger in-vivo primate brain neuroimaging datastes, several are accessible as part of the National Chimpanzee Research Center. More broadly, Navarrete and colleagues have shown quantifiable differences between pre- and post-fixed brain volumes and emphasised the need for caution in the suitability of ex-vivo brain collections to provide reliable volumetric measurements for comparative primate neuroanatomy.
Reardon and colleagues recently published a study on the variation of human brain organization and its relationship with brain size. Using neuroimaging data from more than 3000 individuals they calculated the local surface area and estimated the areal scaling in relation to the total cortical area in order to generate a reference map for areal scaling in cortical and subcortical structures. By using three separate cohorts, three different platforms of image-acquisition, and two distinct imaging processing pipelines, they obtained the same results. Regions with positive scaling, i.e. which area increases with increasing total cortical size, were found with the prefrontal, lateral temporoparietal, and medial parietal cortices, whereas the limbic, primary visual, and primary somatosensory regions showed negative scaling. These patterns of cortical area distribution relatively to normative brain size variation were also reproduced at the individual level in terms of proportion, as, for instance, areas of positive scaling regions were positively correlated with the total cortical area. These patterns of areal scaling distribution are also comparable with patterns of brain expansion during human development and primate evolution (humans vs. macaques). In terms of cytoarchitecture, the regions of positive scaling were concentrated within association cortices, such as the default mode, dorsal attention, and frontoparietal networks, while the negative scaling regions were found within the limbic network. The association of areal scaling patterns with known patterns of mitochondria-related gene expression suggests these regions that are expanded in larger brains might differ in their metabolic profile. The authors concluded that the similarity of the areal scaling maps across development and evolution, and at the individual level, suggests a shared scaling gradient of the primate cortex. Larger brains tend to preferentially expand association cortex, specialized for integration of information, which might point to a need for an increase of the neural subtracts, such as dendrites or synapses, in order to maintain or enhance brain function in an expanded brain. Further study designs are required to investigate the relationship between cortical areal patterns and brain function.
Paleoneurology Lab at Atapuerca, July 2018
Traumatic brain injury (TBI) is a frequent cause of death and disability affecting millions of people every year worldwide. It is initiated by mechanical forces that cause sudden head motion. Such motion produces deformation of the brain and surrounding tissues and thus may result in axonal injury, contusion, or hematoma. The trauma launches a cascade of biochemical reactions often leading to ischemia, hypoxia, brain swelling, and edema. TBI may also induce damages of the cranial vasculature, with alterations of the blood vessel that put the neural tissue at risk. TBI can either cause vessel rupture and hemorrhage (bleeding), or a pathophysiological change of the vessel structure which is secondary associated with some kind of dysfunction. Hemorrhage is easy to recognize, commonly categorized according to its location as epidural (between the skull and the dura mater – associated with disruption of the middle meningeal vessels), subdural (between the dura mater and arachnoid membrane – usually concerning the bridging veins), or subarachnoidal (between arachnoid membrane and the pia mater). Intracerebral bleeding may also occur when the membranes surrounding the brain are impaired. In case of contusion, vascular damage and mechanical cortical damage can occurr at the same time. Even if bleeding is not present, function and microstructure of the injured vessels might be impaired. Vessel disruption and hemorrhage alter the cerebral blood flow, increase the intracranial pressure, affect the maintenance of the blood-brain barrier (exchange of nutrients and waste that occurs at the capillary level), and disrupt the CNS homeostasis by exposing the neural tissue to disregulated blood flow.
Understanding cerebrovascular injuries and the mechanisms behind them is crucial for diagnostics and treatment strategies. Monson et al. (in press) describe the current state of knowledge on the mechanics of cerebral vessels during head trauma and how they respond to the applied loads. They provide a summary of the experimental research focused on the loading conditions during the TBI. Experiments with physical models for instance show that there is significant relative motion between the brain and skull during the trauma. In the sagittal plane, this motion tends to be largest at the vertex and smallest at the brain base. Constraints at the base can lead to brain rotation which pushes the parietal cortex into the cranial bones, possibly causing of contusion and subdural hematoma. In general, it appears that rotation is more damaging than translation. Computer models represent another approach of the TBI research and provide accurate predictions of brain deformation for many loading scenarios. These models enable to estimate exterior loading conditions according to the internal deformation of tissues that are directly involved in injury. However, validation of these models is needed, as well as specific models focusing on blood vessels. The authors also provide a summary of what is known about cerebral vessel response to extreme deformations, passive physical properties, structural failure, and subfailure damage and dysfunction. In another article, Saad et al. (in press) describe various kinds of intracranial hemorrhage in biomedical imaging. Both macroscopic hematomas and microhemorrhages are described according to the distinctions based on intracranial compartments, and traumatic vs nontraumatic cases.
Precise computational modelling of the brain-skull interface is necessary for the prediction, prevention and treatment of acquired brain-injuries. The brain-skull interface comprises complex layers including the osseous cranial tissues, meninges, sub-arachnoidal space and tissues, cerebrospinal fluid (CSF), pia mater and the gray and white cerebral matter. While the tissue properties of the brain-skull interface are known, there is no consensus on how these layers interact during head impact. To generate computational models of the brain-skull interface with greater accuracy, knowing the boundary conditions or constraints is necessary. Previous experimental studies have relied on modelling the deformation of the brain-skull interface using neural density targets (NDTs) implanted into the cadaver brain, collecting information on tissue displacement during front and rear impact in motor vehicle crash-tests.
Wang et al. (2018) utilized computational bio-mechanics and finite element analysis (FEA), placing nodes in the 3D model in close approximation to the position of the experimental NDTs. Four hypotheses of the brain-skull interface were modeled, each approach placing different boundary conditions to model deformation during simulated head impact. All analyses were validated against previous experimental studies. Results showed that how the brain-skull interface was modeled appreciably affected the results. The 3D model showing the closest agreement with the experimental data, included all tissues of the brain-skull interface, allowed for displacement without separation of the skull and brain tissues, and strongly corresponded with known neuroanatomy. This 3D model indicated that non-linear stress-strain associations between brain and skull tissues best matched experimental results. Further, this 3D model could be closely predicted using an Ogden Hyperviscoelastic Constitutive model which did not over- or under-estimate deformations during head impact. The risks of over- and under-estimating head impact during motor vehicle accidents has implications for vehicle construction and prevention of serious brain trauma during accidents. Ultimately, a better understanding of the interaction between layers of the brain-skull interface can produce more accurate predictions of the likely impact during motor vehicle accidents and prevent violent head injury. Extrapolation of this research into paleoneurology could allow investigations into the structural interaction between the brain and braincase, testing if the resistence of brain-skull tissues during deformation evolved in human species as primary adaptations or secondary adjustments such as allometric responses.
Endocranial casts are usually the only resource for studying brain gross anatomy of an extinct species. However, sometimes, a frozen mummy can add information not only on the cortical features but also on the internal structures of the brain. Some years ago, Anastasia Kharlamova and Paul Manger published a study on the mummified brain of a Pleistocene woolly mammoth. This Yuka woolly mammoth specimen, dated approximately to 38,000 years ago, was determined to be an adolescent female aged between 6 and 9 years. Its mummified brain is unique in the state of preservation, allowing access to its external and internal morphology through the use of CT imaging techniques. Furthermore, it gave the opportunity to compare mammoths with extant African elephants’ brain, in order to characterize Elephantidae brain evolution, by determining whether elephant-specific features were already present in this specimen. The authors first compared the brains of the two species in terms of overall organization, concluding the Yuka mammoth displays the typical structure of the Elephantidae family. Direct volume comparisons were possible only for the structures which borders were clearly visible on the CT scans. The brain volume of the Yuka mammoth shrank during mummification, due to dehydration, and occupies only about 55% of the endocranial volume. Therefore, the calculated structure masses had to be corrected for such shrinkage. Moreover, due to differences in tissue fat composition, this shrinkage was heterogeneous, differing between hemispheres and between these and the cerebellum, which showed the least shrinkage. Based on subdural volume and on regression equations using extant mammals, the brain mass was determined to range between 4,230 and 4,340g. Given the average brain mass of an adult female elephant is only 300 to 400 g heavier, and that this specimen is immature, the values appear to be close to what would be expected for an adolescent woolly mammoth female. The size of the corpus callosum was also similar to that of female African and Asian elephants suggesting that, like elephants, mammoths also displayed sexual dimorphism in this structure. The comparable size of the amygdala suggests a similar organization of the limbic system, and the similarity in size and organization of the cerebellum point to a similar role in control of the trunk. This further indicates the Elephantidae family holds the largest cerebellum of all mammals, and that the cerebellar sensorimotor integration and learning movements of the trunk is a feature of this family. As the Elephantidae brain structure seems to be evolutionarily conservative, it can be assumed that the woolly mammoth could have achieved the same cognitive capacities as the extant elephants. However, further predictions of behavior and specializations would need a more detailed histological examination, which was not possible in the Yuka specimen. Nonetheless, this study provided an exceptional glimpse into the brain of an extinct species, and helped extending the understanding of the Elephantidae family.