No abstract

Deep neural networks show high accuracy in the problem of semantic and instance segmentation of biomedical data. However, this approach is computationally expensive. The computational cost may be reduced with network simplification after training or choosing the proper architecture, which provides segmentation with less accuracy but does it much faster. In the present study, we analyzed the accuracy and performance of UNet and ENet architectures for the problem of semantic image segmentation. In addition, we investigated the ENet architecture by replacing of some convolution layers with box-convolution layers. The analysis performed on the original dataset consisted of histology slices with mast cells. These cells provide a region for segmentation with different types of borders, which vary from clearly visible to ragged. ENet was less accurate than UNet by only about 1-2%, but ENet performance was 8-15 times faster than UNet one.

Age-related changes in human cardiomyocytes are closely related to cardiac diseases, especially atrial fibrillation. Restricted availability of biological preparations from the human atrial myocardium complicates experimental studies on the aging processes in cardiomyocytes. In this preliminary study, we used available experimental data on the age-related changes in ionic conductances in canine atrial cardiomyocytes to predict possible consequences of similar remodeling in humans using two mathematical models (Courtemanche98 and Maleckar09) of human atrial cardiomyocytes. The study was performed using the model population approach, allowing one to assess variability in the cellular response to different interventions affecting model parameters. Here, this approach was used to evaluate the effects of age-related parameter modulation on action potential biomarkers in the two models. Simulation results show a significant decrease in the action potential duration and membrane potential at 20% of the action potential duration in aging. These model predictions are consistent with experimental data from mammalians. The action potential characteristics are shown to serve as notable biomarkers of age-related electrophysiological remodeling in human atrial cardiomyocytes. A comparison of the two models shows different behavior in the prediction of repolarization abnormalities.

Electrocardiogram is a widespread method of diagnosis of heart diseases. Nevertheless, there are still issues related to connection of some physiological features of themyocardium with patterns observed on the electrocardiogram. In ourworkwe studied the effect of ventricular remodelling, i.e., thickening ofwalls of ventricles typical for hypertrophic cardiomyopathy (HCM), on the pseudo-electrocardiogram on the surface of a volume conductor during myocardial activation from different sources. A model of two ventricles of the heart was developed for this purpose allowing us to vary ventricular geometry. The volume conductor surrounding the heart was a cubic homogeneous volume conductor. Simulation of a pseudo-electrocardiogram was performed by using a realistic ionic model of cardiomyocytes of the ventricles of the human heart and the bidomain model of the myocardium [15]. The zone of initial activation in the model was given on a part of the subendocardial surface or at one or two points corresponding to positions of electrodes of most common implantable devices. In the course of the study we revealed an inversion of the T-wave when changing the thickness of the left ventricle wall regardless of changes of properties of cardiomyocytes or myocardium conductivity. A linear dependence between the wall thickness of the left ventricle and peak amplitudes and integrals under QRS complex and T wave of the electrocardiogram was shown. We have qualitatively shown that with a change in the wall thickness of the left ventricle the pseudo-electrocardiogram changes stronger in the case of activation from one point than in activation from two points or activation of the entire subendocardium.