VSG-GAN: A high-fidelity image synthesis method with semantic manipulation in retinal fundus image.

In recent years, advancements in retinal image analysis, driven by machine learning and deep learning techniques, have enhanced disease detection and diagnosis through automated feature extraction. However, challenges persist, including limited data set diversity due to privacy concerns and imbalanced sample pairs, hindering effective model training. To address these issues, we introduce the vessel and style guided generative adversarial network (VSG-GAN), an innovative algorithm building upon the foundational concept of GAN. In VSG-GAN, a generator and discriminator engage in an adversarial process to produce realistic retinal images. Our approach decouples retinal image generation into distinct modules: the vascular skeleton and background style. Leveraging style transformation and GAN inversion, our proposed hierarchical variational autoencoder module generates retinal images with diverse morphological traits. In addition, the spatially adaptive denormalization module ensures consistency between input and generated images. We evaluate our model on MESSIDOR and RITE data sets using various metrics, including structural similarity index measure, inception score, Fréchet inception distance, and kernel inception distance. Our results demonstrate the superiority of VSG-GAN, outperforming existing methods across all evaluation assessments. This underscores its effectiveness in addressing data set limitations and imbalances. Our algorithm provides a novel solution to challenges in retinal image analysis by offering diverse and realistic retinal image generation. Implementing the VSG-GAN augmentation approach on downstream diabetic retinopathy classification tasks has shown enhanced disease diagnosis accuracy, further advancing the utility of machine learning in this domain.

Cellular communication among smooth muscle cells: The role of membrane potential via connexins.

Communication via action potentials among neurons has been extensively studied. However, effective communication without action potentials is ubiquitous in biological systems, yet it has received much less attention in comparison. Multi-cellular communication among smooth muscles is crucial for regulating blood flow, for example. Understanding the mechanism of this non-action potential communication is critical in many cases, like synchronization of cellular activity, under normal and pathological conditions. In this paper, we employ a multi-scale asymptotic method to derive a macroscopic homogenized bidomain model from the microscopic electro-neutral (EN) model. This is achieved by considering different diffusion coefficients and incorporating nonlinear interface conditions. Subsequently, the homogenized macroscopic model is used to investigate communication in multi-cellular tissues. Our computational simulations reveal that the membrane potential of syncytia, formed by interconnected cells via connexins, plays a crucial role in propagating oscillations from one region to another, providing an effective means for fast cellular communication. Statement of Significance: In this study, we investigated cellular communication and ion transport in vascular smooth muscle cells, shedding light on their mechanisms under normal and abnormal conditions. Our research highlights the potential of mathematical models in understanding complex biological systems. We developed effective macroscale electro-neutral bi-domain ion transport models and examined their behavior in response to different stimuli. Our findings revealed the crucial role of connexinmediated membrane potential changes and demonstrated the effectiveness of cellular communication through syncytium membranes. Despite some limitations, our study provides valuable insights into these processes and emphasizes the importance of mathematical modeling in unraveling the complexities of cellular communication and ion transport.

MERCI: a machine learning approach to identifying hydroxychloroquine retinopathy using mfERG.

PURPOSE: Hydroxychloroquine (HCQ) is an anti-inflammatory drug in widespread use for the treatment of systemic auto-immune diseases. Vision loss caused by retinal toxicity is a significant risk associated with long term HCQ therapy. Identifying patients at risk of developing retinal toxicity can help prevent vision loss and improve the quality of life for patients. This paper presents updated reference thresholds and examines the diagnostic accuracy of a machine learning approach for identifying retinal toxicity using the multifocal Electroretinogram (mfERG). METHODS: A retrospective study of patients referred for mfERG testing to detect HCQ retinopathy. A consecutive series of all patients referred to Kensington Vision and Research Centre between August 2017 and July 2020 were considered eligible. Eyes suspect for other ocular pathology including widespread retinal disease and advanced macular pathology unrelated to HCQ or with poor quality mfERG recordings were excluded. All patients received mfERG testing and Ocular Coherence Tomography (OCT) imaging. Presence of HCQ retinopathy was based on ring ratio analysis using clinical reference thresholds established at KVRC coupled with structural features observed on OCT, the clinical reference standard. A Support Vector Machine (SVM) using selected features of the mfERG was trained. Accuracy, sensitivity and specificity are reported. RESULTS: 1463 eyes of 748 patients were included in the study. SVM model performance was assessed on 293 eyes from 265 patients. 55 eyes from 54 patients were identified as demonstrating HCQ retinopathy based on the clinical reference standard, 50 eyes from 49 patients were identified by the SVM. Our SVM achieves an accuracy of 85.3% with a sensitivity of 90.9% and specificity of 84.0%. CONCLUSIONS: Machine learning approaches can be applied to mfERG analysis to identify patients at risk of retinopathy caused by HCQ therapy.

A tridomain model for potassium clearance in optic nerve of Necturus.

Complex fluids flow in complex ways in complex structures. Transport of water and various organic and inorganic molecules in the central nervous system are important in a wide range of biological and medical processes. However, the exact driving mechanisms are often not known. In this work, we investigate flows induced by action potentials in an optic nerve as a prototype of the central nervous system. Different from traditional fluid dynamics problems, flows in biological tissues such as the central nervous system are coupled with ion transport. They are driven by osmosis created by concentration gradient of ionic solutions, which in turn influence the transport of ions. Our mathematical model is based on the known structural and biophysical properties of the experimental system used by the Harvard group Orkand et al. Asymptotic analysis and numerical computation show the significant role of water in convective ion transport. The full model (including water) and the electrodiffusion model (excluding water) are compared in detail to reveal an interesting interplay between water and ion transport. In the full model, convection due to water flow dominates inside the glial domain. This water flow in the glia contributes significantly to the spatial buffering of potassium in the extracellular space. Convection in the extracellular domain does not contribute significantly to spatial buffering. Electrodiffusion is the dominant mechanism for flows confined to the extracellular domain.

A mathematical model for persistent post-CSD vasoconstriction.

Cortical spreading depression (CSD) is the propagation of a relatively slow wave in cortical brain tissue that is linked to a number of pathological conditions such as stroke and migraine. Most of the existing literature investigates the dynamics of short term phenomena such as the depolarization and repolarization of membrane potentials or large ion shifts. Here, we focus on the clinically-relevant hour-long state of neurovascular malfunction in the wake of CSDs. This dysfunctional state involves widespread vasoconstriction and a general disruption of neurovascular coupling. We demonstrate, using a mathematical model, that dissolution of calcium that has aggregated within the mitochondria of vascular smooth muscle cells can drive an hour-long disruption. We model the rate of calcium clearance as well as the dynamical implications on overall blood flow. Based on reaction stoichiometry, we quantify a possible impact of calcium phosphate dissolution on the maintenance of F0F1-ATP synthase activity.

Electric discharge of electrocytes: Modelling, analysis and simulation.

In this paper, we investigate the electric discharge of electrocytes by extending our previous work on the generation of electric potential. We first give a complete formulation of a single cell unit consisting of an electrocyte and a resistor, based on a Poisson-Nernst-Planck (PNP) system with various membrane currents as interfacial conditions for the electrocyte and a Maxwell's model for the resistor. Our previous work can be treated as a special case with an infinite resistor (or open circuit). Using asymptotic analysis, we simplify our PNP system and reduce it to an ordinary differential equation (ODE) based model. Unlike the case of an infinite resistor, our numerical simulations of the new model reveal several distinct features. A finite current is generated, which leads to non-constant electric potentials in the bulk of intracellular and extracellular regions. Furthermore, the current induces an additional action potential (AP) at the non-innervated membrane, contrary to the case of an open circuit where an AP is generated only at the innervated membrane. The voltage drop inside the electrocyte is caused by an internal resistance due to mobile ions. We show that our single cell model can be used as the basis for a system with stacked electrocytes and the total current during the discharge of an electric eel can be estimated by using our model.

Electric potential generation of electrocytes: Modelling, analysis, and computation.

In this paper, we developed a one-dimensional model for electric potential generation of electrocytes in electric eels. The model is based on the Poisson-Nernst-Planck system for ion transport coupled with membrane fluxes including the Hodgkin-Huxley type. Using asymptotic analysis, we derived a simplified zero-dimensional model, which we denote as the membrane model in this paper, as a leading order approximation. Our analysis provides justification for the assumption in membrane models that electric potential is constant in the intracellular space. This is essential to explain the superposition of two membrane potentials that leads to a significant transcellular potential. Numerical simulations are also carried out to support our analytical findings.

A Bidomain Model for Lens Microcirculation.

There exists a large body of research on the lens of the mammalian eye over the past several decades. The objective of this work is to provide a link between the most recent computational models and some of the pioneering work in the 1970s and 80s. We introduce a general nonelectroneutral model to study the microcirculation in the lens of the eye. It describes the steady-state relationships among ion fluxes, between water flow and electric field inside cells, and in the narrow extracellular spaces between cells in the lens. Using asymptotic analysis, we derive a simplified model based on physiological data and compare our results with those in the literature. We show that our simplified model can be reduced further to the first-generation models, whereas our full model is consistent with the most recent computational models. In addition, our simplified model captures in its equations the main features of the full computational models. Our results serve as a useful link intermediate between the computational models and the first-generation analytical models. Simplified models of this sort may be particularly helpful as the roles of similar osmotic pumps of microcirculation are examined in other tissues with narrow extracellular spaces, such as cardiac and skeletal muscle, liver, kidney, epithelia in general, and the narrow extracellular spaces of the central nervous system, the "brain." Simplified models may reveal the general functional plan of these systems before full computational models become feasible and specific.

Spatiotemporal model for depth perception in electric sensing.

Electric sensing involves measuring the voltage changes in an actively generated electric field, enabling an environment to be characterized by its electrical properties. It has been applied in a variety of contexts, from geophysics to biomedical imaging. Some species of fish also use an active electric sense to explore their environment in the dark. One of the primary challenges in such electric sensing involves mapping an environment in three-dimensions using voltage measurements that are limited to a two-dimensional sensor array (i.e. a two-dimensional electric image). In some special cases, the distance of simple objects from the sensor array can be estimated by combining properties of the electric image. Here, we describe a novel algorithm for distance estimation based on a single property of the electric image. Our algorithm can be implemented in two simple ways, involving either different electric field strengths or different sensor thresholds, and is robust to changes in object properties and noise.

Predictive models for diabetes mellitus using machine learning techniques.

BACKGROUND: Diabetes Mellitus is an increasingly prevalent chronic disease characterized by the body's inability to metabolize glucose. The objective of this study was to build an effective predictive model with high sensitivity and selectivity to better identify Canadian patients at risk of having Diabetes Mellitus based on patient demographic data and the laboratory results during their visits to medical facilities. METHODS: Using the most recent records of 13,309 Canadian patients aged between 18 and 90 years, along with their laboratory information (age, sex, fasting blood glucose, body mass index, high-density lipoprotein, triglycerides, blood pressure, and low-density lipoprotein), we built predictive models using Logistic Regression and Gradient Boosting Machine (GBM) techniques. The area under the receiver operating characteristic curve (AROC) was used to evaluate the discriminatory capability of these models. We used the adjusted threshold method and the class weight method to improve sensitivity - the proportion of Diabetes Mellitus patients correctly predicted by the model. We also compared these models to other learning machine techniques such as Decision Tree and Random Forest. RESULTS: The AROC for the proposed GBM model is 84.7% with a sensitivity of 71.6% and the AROC for the proposed Logistic Regression model is 84.0% with a sensitivity of 73.4%. The GBM and Logistic Regression models perform better than the Random Forest and Decision Tree models. CONCLUSIONS: The ability of our model to predict patients with Diabetes using some commonly used lab results is high with satisfactory sensitivity. These models can be built into an online computer program to help physicians in predicting patients with future occurrence of diabetes and providing necessary preventive interventions. The model is developed and validated on the Canadian population which is more specific and powerful to apply on Canadian patients than existing models developed from US or other populations. Fasting blood glucose, body mass index, high-density lipoprotein, and triglycerides were the most important predictors in these models.

Mathematical Model of Cardiovascular and Metabolic Responses to Umbilical Cord Occlusions in Fetal Sheep.

Fetal acidemia during labor is associated with an increased risk of brain injury and lasting neurological deficits. This is in part due to the repetitive occlusions of the umbilical cord (UCO) induced by uterine contractions. Whereas fetal heart rate (FHR) monitoring is widely used clinically, it fails to detect fetal acidemia. Hence, new approaches are needed for early detection of fetal acidemia during labor. We built a mathematical model of the UCO effects on FHR, mean arterial blood pressure (MABP), oxygenation and metabolism. Mimicking fetal experiments, our in silico model reproduces salient features of experimentally observed fetal cardiovascular and metabolic behavior including FHR overshoot, gradual MABP decrease and mixed metabolic and respiratory acidemia during UCO. Combined with statistical analysis, our model provides valuable insight into the labor-like fetal distress and guidance for refining FHR monitoring algorithms to improve detection of fetal acidemia and cardiovascular decompensation.

Numerical Study of Cerebroarterial Hemodynamic Changes Following Carotid Artery Operation: A Comparison Between Multiscale Modeling and Stand-Alone Three-Dimensional Modeling.

Free outflow boundary conditions have been widely adopted in hemodynamic model studies, they, however, intrinsically lack the ability to account for the regulatory mechanisms of systemic hemodynamics and hence carry a risk of producing incorrect results when applied to vascular segments with multiple outlets. In the present study, we developed a multiscale model capable of incorporating global cardiovascular properties into the simulation of blood flows in local vascular segments. The multiscale model was constructed by coupling a three-dimensional (3D) model of local arterial segments with a zero-one-dimensional (0-1-D) model of the cardiovascular system. Numerical validation based on an idealized model demonstrated the ability of the multiscale model to preserve reasonable pressure/flow wave transmission among different models. The multiscale model was further calibrated with clinical data to simulate cerebroarterial hemodynamics in a patient undergoing carotid artery operation. The results showed pronounced hemodynamic changes in the cerebral circulation following the operation. Additional numerical experiments revealed that a stand-alone 3D model with free outflow conditions failed to reproduce the results obtained by the multiscale model. These results demonstrated the potential advantage of multiscale modeling over single-scale modeling in patient-specific hemodynamic studies. Due to the fact that the present study was limited to a single patient, studies on more patients would be required to further confirm the findings.

High-throughput single-cell assay for precise measurement of the intrinsic mechanical properties and shape characteristics of red blood cells.

The intrinsic physical and mechanical properties of red blood cells (RBCs), including their geometric and rheological characteristics, can undergo changes in various circulatory and metabolic diseases. However, clinical diagnosis using RBC biophysical phenotypes remains impractical due to the unique biconcave shape, remarkable deformability, and high heterogeneity within different subpopulations. Here, we combine the hydrodynamic mechanisms of fluid-cell interactions in micro circular tubes with a machine learning method to develop a relatively high-throughput microfluidic technology that can accurately measure the shear modulus of the membrane, viscosity, surface area, and volume of individual RBCs. The present method can detect the subtle changes of mechanical properties in various RBC components at continuum scales in response to different doses of cytoskeletal drugs. We also investigate the correlation between glycosylated hemoglobin and RBC mechanical properties. Our study develops a methodology that combines microfluidic technology and machine learning to explore the material properties of cells based on fluid-cell interactions. This approach holds promise in offering novel label-free single-cell-assay-based biophysical markers for RBCs, thereby enhancing the potential for more robust disease diagnosis.

Mathematical approaches to modeling of cortical spreading depression.

Migraine with aura (MwA) is a debilitating disease that afflicts about 25%-30% of migraine sufferers. During MwA, a visual illusion propagates in the visual field, then disappears, and is followed by a sustained headache. MwA was conjectured by Lashley to be related to some neurological phenomenon. A few years later, Leão observed electrophysiological waves in the brain that are now known as cortical spreading depression (CSD). CSD waves were soon conjectured to be the neurological phenomenon underlying MwA that had been suggested by Lashley. However, the confirmation of the link between MwA and CSD was not made until 2001 by Hadjikhani et al. [Proc. Natl. Acad. Sci. U.S.A. 98, 4687-4692 (2001)] using functional MRI techniques. Despite the fact that CSD has been studied continuously since its discovery in 1944, our detailed understandings of the interactions between the mechanisms underlying CSD waves have remained elusive. The connection between MwA and CSD makes the understanding of CSD even more compelling and urgent. In addition to all of the information gleaned from the many experimental studies on CSD since its discovery, mathematical modeling studies provide a general and in some sense more precise alternative method for exploring a variety of mechanisms, which may be important to develop a comprehensive picture of the diverse mechanisms leading to CSD wave instigation and propagation. Some of the mechanisms that are believed to be important include ion diffusion, membrane ionic currents, osmotic effects, spatial buffering, neurotransmitter substances, gap junctions, metabolic pumps, and synaptic connections. Discrete and continuum models of CSD consist of coupled nonlinear differential equations for the ion concentrations. In this review of the current quantitative understanding of CSD, we focus on these modeling paradigms and various mechanisms that are felt to be important for CSD.

Coupled chemical reactions: Effects of electric field, diffusion, and boundary control.

Chemical reactions involve the movement of charges, and this paper presents a mathematical model for describing chemical reactions in electrolytes. The model is developed using an energy variational method that aligns with classical thermodynamics principles. It encompasses both electrostatics and chemical reactions within consistently defined energetic and dissipative functionals. Furthermore, the energy variation method is extended to account for open systems that involve the input and output of charge and mass. Such open systems have the capability to convert one form of input energy into another form of output energy. In particular, a two-domain model is developed to study a reaction system with self-regulation and internal switching, which plays a vital role in the electron transport chain of mitochondria responsible for ATP generation-a crucial process for sustaining life. Simulations are conducted to explore the influence of electric potential on reaction rates and switching dynamics within the two-domain system. It shows that the electric potential inhibits the oxidation reaction while accelerating the reduction reaction.

Analysis of clinical predictors of kidney diseases in type 2 diabetes patients based on machine learning.

BACKGROUND: The heterogeneity of Type 2 Diabetes Mellitus (T2DM) complicated with renal diseases has not been fully understood in clinical practice. The purpose of the study was to propose potential predictive factors to identify diabetic kidney disease (DKD), nondiabetic kidney disease (NDKD), and DKD superimposed on NDKD (DKD + NDKD) in T2DM patients noninvasively and accurately. METHODS: Two hundred forty-one eligible patients confirmed by renal biopsy were enrolled in this retrospective, analytical study. The features composed of clinical and biochemical data prior to renal biopsy were extracted from patients' electronic medical records. Machine learning algorithms were used to distinguish among different kidney diseases pairwise. Feature variables selected in the developed model were evaluated. RESULTS: Logistic regression model achieved an accuracy of 0.8306 ± 0.0057 for DKD and NDKD classification. Hematocrit, diabetic retinopathy (DR), hematuria, platelet distribution width and history of hypertension were identified as important risk factors. Then SVM model allowed us to differentiate NDKD from DKD + NDKD with accuracy 0.8686 ± 0.052 where hematuria, diabetes duration, international normalized ratio (INR), D-Dimer, high-density lipoprotein cholesterol were the top risk factors. Finally, the logistic regression model indicated that DD-dimer, hematuria, INR, systolic pressure, DR were likely to be predictive factors to identify DKD with DKD + NDKD. CONCLUSION: Predictive factors were successfully identified among different renal diseases in type 2 diabetes patients via machine learning methods. More attention should be paid on the coagulation factors in the DKD + NDKD patients, which might indicate a hypercoagulable state and an increased risk of thrombosis.

Multi-objective learning and explanation for stroke risk assessment in Shanxi province.

Stroke is the leading cause of death in China (Zhou et al. in The Lancet, 2019). A dataset from Shanxi Province is analyzed to predict the risk of patients at four states (low/medium/high/attack) and to estimate transition probabilities between various states via a SHAP DeepExplainer. To handle the issues related to an imbalanced sample set, the quadratic interactive deep model (QIDeep) was first proposed by flexible selection and appending of quadratic interactive features. The experimental results showed that the QIDeep model with 3 interactive features achieved the state-of-the-art accuracy 83.33%(95% CI (83.14%; 83.52%)). Blood pressure, physical inactivity, smoking, weight, and total cholesterol are the top five most important features. For the sake of high recall in the attack state, stroke occurrence prediction is considered an auxiliary objective in multi-objective learning. The prediction accuracy was improved, while the recall of the attack state was increased by 17.79% (to 82.06%) compared to QIDeep (from 71.49%) with the same features. The prediction model and analysis tool in this paper provided not only a prediction method but also an attribution explanation of the risk states and transition direction of each patient, a valuable tool for doctors to analyze and diagnose the disease.

Restricted diffusion in cellular media: (1+1)-dimensional model.

We consider the diffusion of molecules in a one-dimensional medium consisting of a large number of cells separated from the extra-cellular space by permeable membranes. The extra-cellular space is completely connected and allows unrestricted diffusion of the molecules. Furthermore, the molecules can diffuse within a given cell, i.e., the intra-cellular space; however, direct diffusion from one cell to another cell cannot occur. There is a movement of molecules across the permeable membranes between the intra- and extra-cellular spaces. Molecules from one cell can cross the permeable membrane into the extra-cellular space, then diffuse through the extra-cellular space, and eventually enter the intra-cellular space of a second cell. Here, we develop a simple set of model equations to describe this phenomenon and obtain the solutions using an eigenfunction expansion. We show that the solutions obtained using this method are particularly convenient for interpreting data from experiments that use techniques from nuclear magnetic resonance imaging.

A continuum neuronal model for the instigation and propagation of cortical spreading depression.

Cortical spreading depression (CSD) waves can occur in the cortices of various brain structures and are associated with the spread of depression of the electroencephalogram signal. In this paper, we present a continuum neuronal model for the instigation and spreading of CSD. Our model assumes that the brain-cell microenvironment can be treated as a porous medium consisting of extra- and intracellular compartments. The main mechanisms in our model for the transport of ions into and out of neurons are cross-membrane ionic currents and (active) pumps, coupled with diffusion in the extracellular space. To demonstrate the applicability of our model, we have carried out extensive numerical simulations under different initial conditions and inclusion of various mechanisms. Our results show that CSD waves can be instigated by injecting cross-membrane ionic currents or by applying KCl in the extracellular space. Furthermore, the estimated speeds of CSD waves are within the experimentally observed range. Effects of specific ion channels, background ion concentrations, extracellular volume fractions, and cell swelling on the propagation speed of CSD are also investigated.

Numerical method for parameter inference of systems of nonlinear ordinary differential equations with partial observations.

Parameter inference of dynamical systems is a challenging task faced by many researchers and practitioners across various fields. In many applications, it is common that only limited variables are observable. In this paper, we propose a method for parameter inference of a system of nonlinear coupled ordinary differential equations with partial observations. Our method combines fast Gaussian process-based gradient matching and deterministic optimization algorithms. By using initial values obtained by Bayesian steps with low sampling numbers, our deterministic optimization algorithm is both accurate, robust and efficient with partial observations and large noise.