Library

Description: The cartilaginous endoskeletons of sharks and rays are covered by tiles of mineralized cartilage called tesserae that enclose areas of unmineralized cartilage. These tesselated layers are vital to the growth as well as the material properties of the skeleton, providing both flexibility and strength. An understanding of the principles behind the tiling of the mineralized layer requires a quantitative analysis of shark and ray skeletal tessellation. However, since a single skeletal element comprises several thousand tesserae, manual segmentation is infeasible. We developed an automated segmentation pipeline that, working from micro-CT data, allows quantification of all tesserae in a skeletal element in less than an hour. Our segmentation algorithm relies on aspects we have learned of general tesseral morphology. In micro-CT scans, tesserae usually appear as round or star-shaped plate-like tiles, wider than deep and connected by mineralized intertesseral joints. Based on these observations, we exploit the distance map of the mineralized layer to separate individual tiles using a hierarchical watershed algorithm. Utilizing a two-dimensional distance map that measures the distance in the plane of the mineralized layer only greatly improves the segmentation. We developed post-processing techniques to quickly correct segmentation errors in regions where tesseral shape differs from the assumed shape. Evaluation of our results is done qualitatively by visual comparison with raw datasets, and quantitatively by comparison to manual segmentations. Furthermore, we generate two-dimensional abstractions of the tiling network based on the neighborhood, allowing representation of complex, biological forms as simpler geometries. We apply our newly developed techniques to the analysis of the left and right hyomandibulae of four ages of stingray enabling the first quantitative analyses of the tesseral tiling structure, while clarifying how these patterns develop across ontogeny.

Description: The endoskeletons of sharks and rays are composed of an unmineralized cartilaginous core, covered in an outer layer of mineralized tiles called tesserae. The tessellated layer is vital to the growth as well as the material properties of the skeletal element, providing both flexibility and strength. However, characterizing the relationship between tesseral size and shape, and skeletal growth and mechanics is challenging because tesserae are small (a few hundred micrometers wide), anchored to the surrounding tissue in complex three-dimensional ways, and occur in huge numbers. Using a custom-made semi-automatic segmentation algorithm, we present the first quantitative and three-dimensional description of tesserae in micro-CT scans of whole skeletal elements. Our segmentation algorithm relies on aspects we have learned of general tesseral morphology. We exploit the distance map of the mineralized layer to separate individual tiles using a hierarchical watershed algorithm. Additionally, we have developed post-processing techniques to quickly correct segmentation errors. Our data reveals that the tessellation is not regular, with tesserae showing a great range of shapes, sizes and number of neighbors. This is partly region-dependent: for example, thick, columnar tesserae are arranged in series along convex edges with small radius of curvature (RoC), whereas more brick-or disc-shaped tesserae are found in planar areas. We apply our newly developed techniques on the left and right hyomandibula (skeletal elements supporting the jaws) from four different ages of a stingray species, to clarify how tiling patterns develop across ontogeny and differ within and between individuals. We evaluate the functional consequences of tesseral morphologies using finite element analysis and 3d-printing, for a better understanding of shark skeletal mechanics, but also to extract fundamental engineering design principles of tiling arrangements on load-bearing three-dimensional objects.

Description: SpiNNaker is a digital, neuromorphic architecture designed for simulating large-scale spiking neural networks at speeds close to biological real-time. Rather than using bespoke analog or digital hardware, the basic computational unit of a SpiNNaker system is a general-purpose ARM processor, allowing it to be programmed to simulate a wide variety of neuron and synapse models. This flexibility is particularly valuable in the study of biological plasticity phenomena. A recently proposed learning rule based on the Bayesian Confidence Propagation Neural Network (BCPNN) paradigm offers a generic framework for modeling the interaction of different plasticity mechanisms using spiking neurons. However, it can be computationally expensive to simulate large networks with BCPNN learning since it requires multiple state variables for each synapse, each of which needs to be updated every simulation time-step. We discuss the trade-offs in efficiency and accuracy involved in developing an event-based BCPNN implementation for SpiNNaker based on an analytical solution to the BCPNN equations, and detail the steps taken to fit this within the limited computational and memory resources of the SpiNNaker architecture. We demonstrate this learning rule by learning temporal sequences of neural activity within a recurrent attractor network which we simulate at scales of up to 2.0 × 10^4 neurons and 5.1 × 10^7 plastic synapses: the largest plastic neural network ever to be simulated on neuromorphic hardware. We also run a comparable simulation on a Cray XC-30 supercomputer system and find that, if it is to match the run-time of our SpiNNaker simulation, the super computer system uses approximately 45× more power. This suggests that cheaper, more power efficient neuromorphic systems are becoming useful discovery tools in the study of plasticity in large-scale brain models.

Description: The cartilaginous endoskeletons of Elasmobranchs (sharks and rays) are reinforced superficially by minute, mineralized tiles, called tesserae. Unlike the bony skeletons of other vertebrates, elasmobranch skeletons have limited healing capability and their tissues’ mechanisms for avoiding damage or managing it when it does occur are largely unknown. Here we describe an aberrant type of mineralized elasmobranch skeletal tissue called endophytic masses (EPMs), which grow into the uncalcified cartilage of the skeleton, but exhibit a strikingly different morphology compared to tesserae and other elasmobranch calcified tissues. We use biological and materials characterization techniques, including computed tomography, electron and light microscopy, x-ray and Raman spectroscopy and histology to characterize the morphology, ultrastructure and chemical composition of tesserae-associated EPMs in different elasmobranch species. EPMs appear to develop between and in intimate association with tesserae, but lack the lines of periodic growth and varying mineral density characteristic of tesserae. EPMs are mineral-dominated (high mineral and low organic content), comprised of birefringent bundles of large monetite or brushite crystals aligned end to end in long strings. Both Unusual skeletal mineralization in elasmobranchs tesserae and EPMs appear to develop in a type-2 collagen-based matrix, but in contrast to tesserae, all chondrocytes embedded or in contact with EPMs are dead and mineralized. The differences outlined between EPMs and tesserae demonstrate them to be distinct tissues. We discuss several possible reasons for EPM development, including tissue reinforcement, repair, and disruptions of mineralization processes, within the context of elasmobranch skeletal biology as well as descriptions of damage responses of other vertebrate mineralized tissues.

Description: The endoskeleton of sharks and rays (elasmobranchs) is comprised of a cartilaginous core, covered by thousands of mineralized tiles, called tesserae. Characterizing the relationship between tesseral morphometrics, skeletal growth and mechanics is challenging because tesserae are small (a few hundred micrometers wide), anchored to the surrounding tissue in complex three-dimensional ways, and occur in huge numbers. We integrate material property, histology, electron microscopy and synchrotron and laboratory µCT scans of skeletal elements from an ontogenetic series of round stingray Urobatis halleri, to gain insights into the generation and maintenance of a natural tessellated system. Using a custom-made semiautomatic segmentation algorithm, we present the first quantitative and 3d description of tesserae across whole skeletal elements. The tessellation is not interlocking or regular, with tesserae showing a great range of shapes, sizes and number of neighbors. This is partly region-dependent: for example, thick, columnar tesserae are arranged in series along convex edges with small radius of curvature (RoC), whereas more brick- or disc-shaped tesserae are found in planar/flatter areas. Comparison of the tessellation across ontogeny, shows that in younger animals, the forming tesseral network is less densely packed, appearing as a covering of separate, poorly mineralized islands that grow together with age to form a complete surface. Some gaps in the tessellation are localized to specific regions in all samples, indicating they are real features, perhaps either regions of delayed mineralization or of tendon insertion. We will use the structure of elasmobranch skeletons as a road map for understanding shark and ray skeletal mechanics, but also to extract fundamental engineering principles for tiled composite materials.

Description: Various 3D imaging techniques are routinely used to examine biological materials, the results of which are usually a stack of grayscale images. In order to quantify structural aspects of the biological materials, however, they must first be extracted from the dataset in a process called segmentation. If the individual structures to be extracted are in contact or very close to each other, distance-based segmentation methods utilizing the Euclidean distance transform are commonly employed. Major disadvantages of the Euclidean distance transform, however, are its susceptibility to noise (very common in biological data), which often leads to incorrect segmentations (i.e. poor separation of objects of interest), and its limitation of being only effective for roundish objects. In the present work, we propose an alternative distance transform method, the random-walk distance transform, and demonstrate its effectiveness in high-throughput segmentation of three microCT datasets of biological tilings (i.e. structures composed of a large number of similar repeating units). In contrast to the Euclidean distance transform, this random-walk approach represents the global, rather than the local, geometric character of the objects to be segmented and, thus, is less susceptible to noise. In addition, it is directly applicable to structures with anisotropic shape characteristics. Using three case studies—stingray tessellated cartilage, starfish dermal endoskeleton, and the prismatic layer of bivalve mollusc shell—we provide a typical workflow for the segmentation of tiled structures, describe core image processing concepts that are underused in biological research, and show that for each study system, large amounts of biologically-relevant data can be rapidly segmented, visualized and analyzed.

Description: Various 3D imaging techniques are routinely used to examine biological materials, the results of which are usually a stack of grayscale images. In order to quantify structural aspects of the biological materials, however, they must first be extracted from the dataset in a process called segmentation. If the individual structures to be extracted are in contact or very close to each other, distance-based segmentation methods utilizing the Euclidean distance transform are commonly employed. Major disadvantages of the Euclidean distance transform, however, are its susceptibility to noise (very common in biological data), which often leads to incorrect segmentations (i.e. poor separation of objects of interest), and its limitation of being only effective for roundish objects. In the present work, we propose an alternative distance transform method, the random-walk distance transform, and demonstrate its effectiveness in high-throughput segmentation of three microCT datasets of biological tilings (i.e. structures composed of a large number of similar repeating units). In contrast to the Euclidean distance transform, this random-walk approach represents the global, rather than the local, geometric character of the objects to be segmented and, thus, is less susceptible to noise. In addition, it is directly applicable to structures with anisotropic shape characteristics. Using three case studies—stingray tessellated cartilage, starfish dermal endoskeleton, and the prismatic layer of bivalve mollusc shell—we provide a typical workflow for the segmentation of tiled structures, describe core image processing concepts that are underused in biological research, and show that for each study system, large amounts of biologically-relevant data can be rapidly segmented, visualized and analyzed.