Identifying and training deep learning neural networks on biomedical-related datasets.

This manuscript describes the development of a resources module that is part of a learning platform named 'NIGMS Sandbox for Cloud-based Learning' https://github.com/NIGMS/NIGMS-Sandbox. The overall genesis of the Sandbox is described in the editorial NIGMS Sandbox at the beginning of this Supplement. This module delivers learning materials on implementing deep learning algorithms for biomedical image data in an interactive format that uses appropriate cloud resources for data access and analyses. Biomedical-related datasets are widely used in both research and clinical settings, but the ability for professionally trained clinicians and researchers to interpret datasets becomes difficult as the size and breadth of these datasets increases. Artificial intelligence, and specifically deep learning neural networks, have recently become an important tool in novel biomedical research. However, use is limited due to their computational requirements and confusion regarding different neural network architectures. The goal of this learning module is to introduce types of deep learning neural networks and cover practices that are commonly used in biomedical research. This module is subdivided into four submodules that cover classification, augmentation, segmentation and regression. Each complementary submodule was written on the Google Cloud Platform and contains detailed code and explanations, as well as quizzes and challenges to facilitate user training. Overall, the goal of this learning module is to enable users to identify and integrate the correct type of neural network with their data while highlighting the ease-of-use of cloud computing for implementing neural networks. This manuscript describes the development of a resource module that is part of a learning platform named ``NIGMS Sandbox for Cloud-based Learning'' https://github.com/NIGMS/NIGMS-Sandbox. The overall genesis of the Sandbox is described in the editorial NIGMS Sandbox [1] at the beginning of this Supplement. This module delivers learning materials on the analysis of bulk and single-cell ATAC-seq data in an interactive format that uses appropriate cloud resources for data access and analyses.

A three-dimensional valve-on-chip microphysiological system implicates cell cycle progression, cholesterol metabolism and protein homeostasis in early calcific aortic valve disease progression.

BACKGROUND: Calcific aortic valve disease (CAVD) is one of the most common forms of valvulopathy, with a 50 % elevated risk of a fatal cardiovascular event, and greater than 15,000 annual deaths in North America alone. The treatment standard is valve replacement as early diagnostic, mitigation, and drug strategies remain underdeveloped. The development of early diagnostic and therapeutic strategies requires the fabrication of effective in vitro valve mimetic models to elucidate early CAVD mechanisms. METHODS: In this study, we developed a multilayered physiologically relevant 3D valve-on-chip (VOC) system that incorporated aortic valve mimetic extracellular matrix (ECM), porcine aortic valve interstitial cell (VIC) and endothelial cell (VEC) co-culture and dynamic mechanical stimuli. Collagen and glycosaminoglycan (GAG) based hydrogels were assembled in a bilayer to mimic healthy or diseased compositions of the native fibrosa and spongiosa. Multiphoton imaging and proteomic analysis of healthy and diseased VOCs were performed. RESULTS: Collagen-based bilayered hydrogel maintained the phenotype of the VICs. Proteins related to cellular processes like cell cycle progression, cholesterol biosynthesis, and protein homeostasis were found to be significantly altered and correlated with changes in cell metabolism in diseased VOCs. This study suggested that diseased VOCs may represent an early, adaptive disease initiation stage, which was corroborated by human aortic valve proteomic assessment. CONCLUSIONS: In this study, we developed a collagen-based bilayered hydrogel to mimic healthy or diseased compositions of the native fibrosa and spongiosa layers. When the gels were assembled in a VOC with VECs and VICs, the diseased VOCs revealed key insights about the CAVD initiation process. STATEMENT OF SIGNIFICANCE: Calcific aortic valve disease (CAVD) elevates the risk of death due to cardiovascular pathophysiology by 50 %, however, prevention and mitigation strategies are lacking, clinically. Developing tools to assess early disease would significantly aid in the prevention of disease and in the development of therapeutics. Previously, studies have utilized collagen and glycosaminoglycan-based hydrogels for valve cell co-cultures, valve cell co-cultures in dynamic environments, and inorganic polymer-based multilayered hydrogels; however, these approaches have not been combined to make a physiologically relevant model for CAVD studies. We fabricated a bi-layered hydrogel that closely mimics the aortic valve and used it for valve cell co-culture in a dynamic platform to gain mechanistic insights into the CAVD initiation process using proteomic and multiphoton imaging assessment.

Pre-Incisional and Multiple Intradermal Injection of N-Acetylcysteine Slightly Improves Incisional Wound Healing in an Animal Model.

The objective of this study was to investigate if delivering multiple doses of N-acetylcysteine (NAC) post-surgery in addition to pre-incisional administration significantly impacts the wound healing process in a rat model. Full-thickness skin incisions were carried out on the dorsum of 24 Sprague-Dawley rats in six locations. Fifteen minutes prior to the incision, half of the sites were treated with a control solution, with the wounds on the contralateral side treated with solutions containing 0.015%, 0.03% and 0.045% of NAC. In the case of the NAC treated group, further injections were given every 8 h for three days. On days 3, 7, 14 and 60 post-op, rats were sacrificed to gather material for the histological analysis, which included histomorphometry, collagen fiber organization analysis, immunohistochemistry and Abramov scale scoring. It was determined that scars treated with 0.015% NAC had significantly lower reepithelization than the control at day 60 post-op (p = 0.0018). Scars treated with 0.045% NAC had a significantly lower collagen fiber variance compared to 0.015% NAC at day 14 post-op (p = 0.02 and p = 0.04) and a lower mean scar width than the control at day 60 post-op (p = 0.0354 and p = 0.0224). No significant differences in the recruitment of immune cells and histological parameters were found. The results point to a limited efficacy of multiple NAC injections post-surgery in wound healing.

Quantification of age-related changes in the structure and mechanical function of skin with multiscale imaging.

The mechanical properties of skin change during aging but the relationships between structure and mechanical function remain poorly understood. Previous work has shown that young skin exhibits a substantial decrease in tissue volume, a large macro-scale Poisson's ratio, and an increase in micro-scale collagen fiber alignment during mechanical stretch. In this study, label-free multiphoton microscopy was used to quantify how the microstructure and fiber kinematics of aged mouse skin affect its mechanical function. In an unloaded state, aged skin was found to have less collagen alignment and more non-enzymatic collagen fiber crosslinks. Skin samples were then loaded in uniaxial tension and aged skin exhibited a lower mechanical stiffness compared to young skin. Aged tissue also demonstrated less volume reduction and a lower macro-scale Poisson's ratio at 10% uniaxial strain, but not at 20% strain. The magnitude of 3D fiber realignment in the direction of loading was not different between age groups, and the amount of realignment in young and aged skin was less than expected based on theoretical fiber kinematics affine to the local deformation. These findings provide key insights on how the collagen fiber microstructure changes with age, and how those changes affect the mechanical function of skin, findings which may help guide wound healing or anti-aging treatments.

Mechanical Models of Collagen Networks for Understanding Changes in the Failure Properties of Aging Skin.

Skin undergoes mechanical alterations due to changes in the composition and structure of the collagenous dermis with aging. Previous studies have conflicting findings, with both increased and decreased stiffness reported for aging skin. The underlying structure-function relationships that drive age-related changes are complex and difficult to study individually. One potential contributor to these variations is the accumulation of nonenzymatic crosslinks within collagen fibers, which affect dermal collagen remodeling and mechanical properties. Specifically, these crosslinks make individual fibers stiffer in their plastic loading region and lead to increased fragmentation of the collagenous network. To better understand the influence of these changes, we investigated the impact of nonenzymatic crosslink changes on the dermal microstructure using discrete fiber networks representative of the dermal microstructure. Our findings suggest that stiffening the plastic region of collagen's mechanical response has minimal effects on network-level stiffness and failure stresses. Conversely, simulating fragmentation through a loss of connectivity substantially reduces network stiffness and failure stress, while increasing stretch ratios at failure.

Neuro-regenerative behavior of adipose-derived stem cells in aligned collagen I hydrogels.

Peripheral nerve injuries persist as a major clinical issue facing the US population and can be caused by stretch, laceration, or crush injuries. Small nerve gaps are simple to treat, and the nerve stumps can be reattached with sutures. In longer nerve gaps, traditional treatment options consist of autografts, hollow nerve guidance conduits, and, more recently, manufactured fibrous scaffolds. These manufactured scaffolds often incorporate stem cells, growth factors, and/or extracellular matrix (ECM) proteins to better mimic the native environment but can have issues with homogenous cell distribution or uniformly oriented neurite outgrowth in scaffolds without fibrous alignment. Here, we utilize a custom device to fabricate collagen I hydrogels with aligned fibers and encapsulated adipose-derived mesenchymal stem cells (ASCs) for potential use as a peripheral nerve repair graft. Initial results of our scaffold system revealed significantly less cell viability in higher collagen gel concentrations; 3 mg/mL gels showed 84.8 ± 7.3% viable cells, compared to 6 mg/mL gels viability of 76.7 ± 9.5%. Mechanical testing of the 3 mg/mL gels showed a Young's modulus of 6.5 ± 0.8 kPa nearly matching 7.45 kPa known to support Schwann cell migration. Further analysis of scaffolds coupled with stretching in vitro revealed heightened angiogenic and factor secretion, ECM deposition, fiber alignment, and dorsal root ganglia (DRG) neurite outgrowth along the axis of fiber alignment. Our platform serves as an in vitro testbed to assess neuro-regenerative potential of ASCs in aligned collagen fiber scaffolds and may provide guidance on next-generation nerve repair scaffold design.

Improved segmentation of collagen second harmonic generation images with a deep learning convolutional neural network.

Collagen fibers play an important role in both the structure and function of various tissues in the human body. Visualization and quantitative measurements of collagen fibers are possible through imaging modalities such as second harmonic generation (SHG), but accurate segmentation of collagen fibers is difficult for datasets involving variable imaging depths due to the effects of scattering and absorption. Therefore, an objective approach to segmentation is needed for datasets with images of variable SHG intensity. In this study, a U-Net convolutional neural network (CNN) was trained to accurately segment collagen-positive pixels throughout SHG z-stacks. CNN performance was benchmarked against other common thresholding techniques, and was found to outperform intensity-based segmentation algorithms within an independent dataset, particularly at deeper imaging depths. These results indicate that a trained CNN can accurately segment collagen-positive pixels within a wide range of imaging depths, which is useful for quantitative SHG imaging in thick tissues.

Multiscale Computational Model Predicts Mouse Skin Kinematics Under Tensile Loading.

Skin is a complex tissue whose biomechanical properties are generally understood in terms of an incompressible material whose microstructure undergoes affine deformations. A growing number of experiments, however, have demonstrated that skin has a high Poisson's ratio, substantially decreases in volume during uniaxial tensile loading, and demonstrates collagen fiber kinematics that are not affine with local deformation. In order to better understand the mechanical basis for these properties, we constructed multiscale mechanical models (MSM) of mouse skin based on microstructural multiphoton microscopy imaging of the dermal microstructure acquired during mechanical testing. Three models that spanned the cases of highly aligned, moderately aligned, and nearly random fiber networks were examined and compared to the data acquired from uniaxially stretched skin. Our results demonstrate that MSMs consisting of networks of matched fiber organization can predict the biomechanical behavior of mouse skin, including the large decrease in tissue volume and nonaffine fiber kinematics observed under uniaxial tension.

Single Dose of N-Acetylcysteine in Local Anesthesia Increases Expression of HIF1α, MAPK1, TGFß1 and Growth Factors in Rat Wound Healing.

In this study, we aimed to investigate the influence of N-acetylcysteine (NAC) on the gene expression profile, neoangiogenesis, neutrophils and macrophages in a rat model of incisional wounds. Before creating wounds on the backs of 24 Sprague-Dawley rats, intradermal injections were made. Lidocaine-epinephrin solutions were supplemented with 0.015%, 0.03% or 0.045% solutions of NAC, or nothing (control group). Scars were harvested on the 3rd, 7th, 14th and 60th day post-surgery. We performed immunohistochemical staining in order to visualize macrophages (anti-CD68), neutrophils (anti-MPO) and newly formed blood vessels (anti-CD31). Additionally, RT-qPCR was used to measure the relative expression of 88 genes involved in the wound healing process. On the 14th day, the number of cells stained with anti-CD68 and anti-CD31 antibodies was significantly larger in the tissues treated with 0.03% NAC compared with the control. Among the selected genes, 52 were upregulated and six were downregulated at different time points. Interestingly, NAC exerted a significant effect on the expression of 45 genes 60 days after its administration. In summation, a 0.03% NAC addition to the pre-incisional anesthetic solution improves neovasculature and increases the macrophages' concentration at the wound site on the 14th day, as well as altering the expression of numerous genes that are responsible for the regenerative processes.

N-Acetylcysteine Added to Local Anesthesia Reduces Scar Area and Width in Early Wound Healing-An Animal Model Study.

The aim of the study was to evaluate if a pre-incisional N-acetylcysteine (NAC) treatment altered the process of wound healing in a rat model. The dorsal skin of 24 Sprague-Dawley rats was incised in six locations. Before the incisions were made, skin was injected either with lidocaine and epinephrine (one side) or with these agents supplemented with 0.015%, 0.03%, or 0.045% NAC (contralaterally). Photographic documentation of the wound healing process was made at 11 time points. Rats were sacrificed 3, 7, 14, or 60 days after incision to excise scars for histological analysis. They included: Abramov scale scoring, histomorphometry analysis, and collagen fiber arrangement assessment. Skin pretreated with 0.03% NAC produced the shortest scars at all analyzed time points, though this result was statistically insignificant. At this NAC concentration the scars had smaller areas on the third day and were narrower on the day 4 compared with all the other groups (p < 0.05). On day 7, at the same concentration of NAC, the scars had a higher superficial concentration index (p = 0.03) and larger dermal proliferation area (p = 0.04). NAC addition to pre-incisional anesthetic solution decreased wound size and width at an early stage of scar formation at all concentrations; however, with optimal results at 0.03% concentration.

Efficacy of Combined in-vivo Electroporation-Mediated Gene Transfer of VEGF, HGF, and IL-10 on Skin Flap Survival, Monitored by Label-Free Optical Imaging: A Feasibility Study.

Preventing surgical flaps necrosis remains challenging. Laser Doppler imaging and ultrasound can monitor blood flow in flap regions, but they do not directly measure the cellular response to ischemia. The study aimed to investigate the efficacy of synergistic in-vivo electroporation-mediated gene transfer of interleukin 10 (IL-10) with either hepatocyte growth factor (HGF) or vascular endothelial growth factor (VEGF) on the survival of a modified McFarlane flap, and to evaluate the effect of the treatment on cell metabolism, using label-free fluorescence lifetime imaging. Fifteen male Wistar rats (290-320 g) were randomly divided in three groups: group-A (control group) underwent surgery and received no gene transfer. Group-B received electroporation mediated hIL-10 gene delivery 24 h before and VEGF gene delivery 24 h after surgery. Group-C received electroporation mediated hIL-10 gene delivery 24 h before and hHGF gene delivery 24 h after surgery. The animals were assessed clinically and histologically. In addition, label-free fluorescence lifetime imaging was performed on the flap. Synergistic electroporation mediated gene delivery significantly decreased flap necrosis (P = 0.0079) and increased mean vessel density (P = 0.0079) in treatment groups B and C compared to control group-A. NADH fluorescence lifetime analysis indicated an increase in oxidative phosphorylation in the epidermis of the group-B (P = 0.039) relative to controls. These findings suggested synergistic in-vivo electroporation-mediated gene transfer as a promising therapeutic approach to enhance viability and vascularity of skin flap. Furthermore, the study showed that combinational gene therapy promoted an increase in tissue perfusion and a relative increase in oxidative metabolism within the epithelium.

Three-Dimensional Quantification of Collagen Microstructure During Tensile Mechanical Loading of Skin.

Skin is a heterogeneous tissue that can undergo substantial structural and functional changes with age, disease, or following injury. Understanding how these changes impact the mechanical properties of skin requires three-dimensional (3D) quantification of the tissue microstructure and its kinematics. The goal of this study was to quantify these structure-function relationships via second harmonic generation (SHG) microscopy of mouse skin under tensile mechanical loading. Tissue deformation at the macro- and micro-scale was quantified, and a substantial decrease in tissue volume and a large Poisson's ratio was detected with stretch, indicating the skin differs substantially from the hyperelastic material models historically used to explain its behavior. Additionally, the relative amount of measured strain did not significantly change between length scales, suggesting that the collagen fiber network is uniformly distributing applied strains. Analysis of undeformed collagen fiber organization and volume fraction revealed a length scale dependency for both metrics. 3D analysis of SHG volumes also showed that collagen fiber alignment increased in the direction of stretch, but fiber volume fraction did not change. Interestingly, 3D fiber kinematics was found to have a non-affine relationship with tissue deformation, and an affine transformation of the micro-scale fiber network overestimates the amount of fiber realignment. This result, along with the other outcomes, highlights the importance of accurate, scale-matched 3D experimental measurements when developing multi-scale models of skin mechanical function.

Quantifying Age-Related Changes in Skin Wound Metabolism Using In Vivo Multiphoton Microscopy.

Objective: The elderly are at high risk for developing chronic skin wounds, but the effects of intrinsic aging on skin healing are difficult to isolate due to common comorbidities like diabetes. Our objective is to use multiphoton microscopy (MPM) to find endogenous, noninvasive biomarkers to differentiate changes in skin wound healing metabolism between young and aged mice in vivo. Approach: We utilized MPM to monitor skin metabolism at the edge of full-thickness, excisional wounds in 24- and 4-month-old mice of both sexes for 10 days. MPM can assess quantitative biomarkers of cellular metabolism in vivo by utilizing autofluorescence from the cofactors nicotinamide adenine dinucleotide (NADH) and flavin adenine dinucleotide (FAD). Results: An optical redox ratio of FAD/(NADH+FAD) autofluorescence and NADH fluorescence lifetime imaging revealed dynamic changes in keratinocyte function during healing. Aged female mice demonstrated an attenuation of keratinocyte proliferation during wound healing detectable optically through a higher redox ratio and longer NADH fluorescence lifetime. By measuring the correlation between NADH lifetime and the optical redox ratio at each day, we also demonstrate sensitivity to the proliferative phase of wound healing. Innovation: Label-free MPM was used to longitudinally monitor individual wounds in vivo, which revealed age-dependent differences in wound metabolism. Conclusion: These results indicate in vivo MPM can provide quantitative biomarkers of age-related delays in healing, which can be used in the future to provide patient-specific wound care.

Skin Structure-Function Relationships and the Wound Healing Response to Intrinsic Aging.

Significance: Chronic wounds, such as diabetic foot ulcers, venous stasis ulcers, and pressure ulcers affect millions of Americans each year, and disproportionately afflict our increasingly older population. Older individuals are predisposed to wound infection, repeated trauma, and the development of chronic wounds. However, a complete understanding of how the attributes of aging skin affect the wound healing process has remained elusive. Recent Advances: A variety of studies have demonstrated that the dermal matrix becomes thinner, increasingly crosslinked, and fragmented with advanced age. These structural changes, as well as an increase in cell senescence, result in altered collagen fiber remodeling and increased stiffness. Studies combining mechanical testing with advanced imaging techniques are providing new insights into the relationships between these age-related changes. Emerging research into the mechanobiology of aging and the wound healing process indicate that the altered mechanical environment of aged skin may have a significant effect on age-related delays in healing. Critical Issues: The interpretation and synthesis of clinical studies is confounded by the effects of common comorbidities that also contribute to the development of chronic wounds. A lack of quantitative biomarkers of wound healing and age-related changes makes understanding structure-function relationships during the wound healing process challenging. Future Directions: Additional work is needed to establish quantitative and mechanistic relationships among age-related changes in the skin microstructure, mechanical function, and the cellular responses to wound healing.

Characterizing differences in the collagen fiber organization of skin wounds using quantitative polarized light imaging.

Collagen fiber organization requires characterization in many biomedical applications, but it is difficult to objectively quantify in standard histology tissue sections. Quantitative polarized light imaging is a low-cost technique that allows for rapid measurement of collagen fiber orientation and thickness. In this study, we utilize a quantitative polarized light imaging system to characterize fiber orientation and thickness from wound sections. Full thickness skin wound sections that were previously stained with hematoxylin and eosin were used to assess collagen fiber content and organization at different points during the wound healing process. Overall, wounds exhibited a measurable increase in collagen fiber thickness and a nonlinear change in fiber reorganization within the wound. Our study demonstrates that quantitative polarized light imaging is an inexpensive alternative or supplement to standard histology protocols, requiring no additional stains or dyes, and yields repeatable quantitative assessments of collagen organization.

In vivo multiphoton microscopy detects longitudinal metabolic changes associated with delayed skin wound healing.

Chronic wounds are difficult to diagnose and characterize due to a lack of quantitative biomarkers. Label-free multiphoton microscopy has emerged as a useful imaging modality capable of quantifying changes in cellular metabolism using an optical redox ratio of FAD/(NADH+FAD) autofluorescence. However, the utility of an optical redox ratio for long-term in vivo monitoring of tissue metabolism has not been robustly evaluated. In this study, we demonstrate how multiphoton microscopy can be used to monitor changes in the metabolism of individual full-thickness skin wounds in vivo. 3D optical redox ratio maps and NADH fluorescence lifetime images identify differences between diabetic and control mice during the re-epithelialization of wounds. These metabolic changes are associated with a transient increase in keratinocyte proliferation at the wound edge. Our study demonstrates that high-resolution, non-invasive autofluorescence imaging can be performed in vivo and that optical redox ratios can serve as quantitative optical biomarkers of impaired wound healing.