Graph Empirical Mode Decomposition-Based Data Augmentation Applied to Gifted Children MRI Analysis.

Gifted children and normal controls can be distinguished by analyzing the structural connectivity (SC) extracted from MRI data. Previous studies have improved classification accuracy by extracting several features of the brain regions. However, the limited size of the database may lead to degradation when training deep neural networks as classification models. To this end, we propose to use a data augmentation method by adding artificial samples generated using graph empirical mode decomposition (GEMD). We decompose the training samples by GEMD to obtain the intrinsic mode functions (IMFs). Then, the IMFs are randomly recombined to generate the new artificial samples. After that, we use the original training samples and the new artificial samples to enlarge the training set. To evaluate the proposed method, we use a deep neural network architecture called BrainNetCNN to classify the SCs of MRI data with and without data augmentation. The results show that the data augmentation with GEMD can improve the average classification performance from 55.7 to 78%, while we get a state-of-the-art classification accuracy of 93.3% by using GEMD in some cases. Our results demonstrate that the proposed GEMD augmentation method can effectively increase the limited number of samples in the gifted children dataset, improving the classification accuracy. We also found that the classification accuracy is improved when specific features extracted from brain regions are used, achieving 93.1% for some feature selection methods.

Network Analysis Reveals TNF as a Major Hub of Reactive Inflammation Following Spinal Cord Injury.

Spinal cord injury (SCI) leads to reactive inflammation and other harmful events that limit spinal cord regeneration. We propose an approach for studying the mechanisms at the levels of network topology, gene ontology, signaling pathways, and disease inference. We treated inflammatory mediators as toxic chemicals and retrieved the genes and interacting proteins associated with them via a set of biological medical databases and software. We identified >10,000 genes associated with SCI. Tumor necrosis factor (TNF) had the highest scores, and the top 30 were adopted as core data. In the core interacting protein network, TNF and other top 10 nodes were the major hubs. The core members were involved in cellular responses and metabolic processes, as components of the extracellular space and regions, in protein-binding and receptor-binding functions, as well as in the TNF signaling pathway. In addition, both seizures and SCI were highly associated with TNF levels; therefore, for achieving a better curative effect on SCI, TNF and other major hubs should be targeted together according to the theory of network intervention, rather than a single target such as TNF alone. Furthermore, certain drugs used to treat epilepsy could be used to treat SCI as adjuvants.

Numerical characterization of regenerative axons growing along a spherical multifunctional scaffold after spinal cord injury.

Spinal cord injury (SCI) followed by extensive cell loss, inflammation, and scarring, often permanently damages neurological function. Biomaterial scaffolds are promising but currently have limited applicability in SCI because after entering the scaffold, regenerating axons tend to become trapped and rarelyre-enter the host tissue, the reasons for which remain to be completely explored. Here, we propose a mathematical model and computer simulation for characterizing regenerative axons growing along a scaffold following SCI, and how their growth may be guided. The model assumed a solid, spherical, multifunctional, biomaterial scaffold, that would bridge the rostral and caudal stumps of a completely transected spinal cord in a rat model and would guide the rostral regenerative axons toward the caudal tissue. Other assumptions include the whole scaffold being coated with extracellular matrix components, and the caudal area being additionally seeded with chemoattractants. The chemical factors on and around the scaffold were formulated to several coupled variables, and the parameter values were derived fromexisting experimental data. Special attention was given to the effects of coating strength, seeding location, and seeding density, as well as the ramp slope of the scaffold, on axonal regeneration. In numerical simulations, a slimmer scaffold provided a small slope at the entry "on-ramp" area that improved the success rate of axonal regeneration. If success rates are high, an increased number of regenerative axons traverse through the narrow channels, causing congestion and lowering the growth rate. An increase in the number of severed axons (300-12000) did not significantly affect the growth rate, but it reduced the success rate of axonal regeneration. However, an increase in the seeding densities of the complexes on the whole scaffold, and that in the seeding densities of the chemoattractants on the caudal area, improved both the success and growth rates. However, an increase in the density of thecomplexes on the whole scaffold risks an over-eutrophic surface that harms axonal regeneration.Although theoretical predictions are yet to be validated directly by experiments, this theoretical tool can advance the treatment of SCI, and is also applicable to scaffolds with other architectures.

A Mathematical Model of Regenerative Axon Growing along Glial Scar after Spinal Cord Injury.

A major factor in the failure of central nervous system (CNS) axon regeneration is the formation of glial scar after the injury of CNS. Glial scar generates a dense barrier which the regenerative axons cannot easily pass through or by. In this paper, a mathematical model was established to explore how the regenerative axons grow along the surface of glial scar or bypass the glial scar. This mathematical model was constructed based on the spinal cord injury (SCI) repair experiments by transplanting Schwann cells as bridge over the glial scar. The Lattice Boltzmann Method (LBM) was used in this model for three-dimensional numerical simulation. The advantage of this model is that it provides a parallel and easily implemented algorithm and has the capability of handling complicated boundaries. Using the simulated data, two significant conclusions were made in this study: (1) the levels of inhibitory factors on the surface of the glial scar are the main factors affecting axon elongation and (2) when the inhibitory factor levels on the surface of the glial scar remain constant, the longitudinal size of the glial scar has greater influence on the average rate of axon growth than the transverse size. These results will provide theoretical guidance and reference for researchers to design efficient experiments.

Glial scar size, inhibitor concentration, and growth of regenerating axons after spinal cord transection.

A mathematical model has been formulated in accordance with cell chemotaxis and relevant experimental data. A three-dimensional lattice Boltzmann method was used for numerical simulation. The present study observed the effects of glial scar size and inhibitor concentration on regenerative axonal growth following spinal cord transection. The simulation test comprised two parts: (1) when release rates of growth inhibitor and promoter were constant, the effects of glial scar size on axonal growth rate were analyzed, and concentrations of inhibitor and promoters located at the moving growth cones were recorded. (2) When the glial scar size was constant, the effects of inhibitor and promoter release rates on axonal growth rate were analyzed, and inhibitor and promoter concentrations at the moving growth cones were recorded. Results demonstrated that (1) a larger glial scar and a higher release rate of inhibitor resulted in a reduced axonal growth rate. (2) The axonal growth rate depended on the ratio of inhibitor to promoter concentrations at the growth cones. When the average ratio was < 1.5, regenerating axons were able to grow and successfully contact target cells.