Constrained and unconstrained deep image prior optimization models with automatic regularization.

Deep Image Prior (DIP) is currently among the most efficient unsupervised deep learning based methods for ill-posed inverse problems in imaging. This novel framework relies on the implicit regularization provided by representing images as the output of generative Convolutional Neural Network (CNN) architectures. So far, DIP has been shown to be an effective approach when combined with classical and novel regularizers. Unfortunately, to obtain appropriate solutions, all the models proposed up to now require an accurate estimate of the regularization parameter. To overcome this difficulty, we consider a locally adapted regularized unconstrained model whose local regularization parameters are automatically estimated for additively separable regularizers. Moreover, we propose a novel constrained formulation in analogy to Morozov's discrepancy principle which enables the application of a broader range of regularizers. Both the unconstrained and the constrained models are solved via the proximal gradient descent-ascent method. Numerical results demonstrate the robustness with respect to image content, noise levels and hyperparameters of the proposed models on both denoising and deblurring of simulated as well as real natural and medical images.

DeepCEL0 for 2D single-molecule localization in fluorescence microscopy.

MOTIVATION: In fluorescence microscopy, single-molecule localization microscopy (SMLM) techniques aim at localizing with high-precision high-density fluorescent molecules by stochastically activating and imaging small subsets of blinking emitters. Super resolution plays an important role in this field since it allows to go beyond the intrinsic light diffraction limit. RESULTS: In this work, we propose a deep learning-based algorithm for precise molecule localization of high-density frames acquired by SMLM techniques whose ℓ2-based loss function is regularized by non-negative and ℓ0-based constraints. The ℓ0 is relaxed through its continuous exact ℓ0 (CEL0) counterpart. The arising approach, named DeepCEL0, is parameter-free, more flexible, faster and provides more precise molecule localization maps if compared to the other state-of-the-art methods. We validate our approach on both simulated and real fluorescence microscopy data. AVAILABILITY AND IMPLEMENTATION: DeepCEL0 code is freely accessible at https://github.com/sedaboni/DeepCEL0.

Treatment With Treprostinil and Metformin Normalizes Hyperglycemia and Improves Cardiac Function in Pulmonary Hypertension Associated With Heart Failure With Preserved Ejection Fraction.

OBJECTIVE: Pulmonary hypertension (PH) due to left heart disease (group 2), especially in the setting of heart failure with preserved ejection fraction (HFpEF), is the most common cause of PH worldwide; however, at present, there is no proven effective therapy available for its treatment. PH-HFpEF is associated with insulin resistance and features of metabolic syndrome. The stable prostacyclin analog, treprostinil, is an effective and widely used Food and Drug Administration-approved drug for the treatment of pulmonary arterial hypertension. While the effect of treprostinil on metabolic syndrome is unknown, a recent study suggests that the prostacyclin analog beraprost can improve glucose intolerance and insulin sensitivity. We sought to evaluate the effectiveness of treprostinil in the treatment of metabolic syndrome-associated PH-HFpEF. Approach and Results: Treprostinil treatment was given to mice with mild metabolic syndrome-associated PH-HFpEF induced by high-fat diet and to SU5416/obese ZSF1 rats, a model created by the treatment of rats with a more profound metabolic syndrome due to double leptin receptor defect (obese ZSF1) with a vascular endothelial growth factor receptor blocker SU5416. In high-fat diet-exposed mice, chronic treatment with treprostinil reduced hyperglycemia and pulmonary hypertension. In SU5416/Obese ZSF1 rats, treprostinil improved hyperglycemia with similar efficacy to that of metformin (a first-line drug for type 2 diabetes mellitus); the glucose-lowering effect of treprostinil was further potentiated by the combined treatment with metformin. Early treatment with treprostinil in SU5416/Obese ZSF1 rats lowered pulmonary pressures, and a late treatment with treprostinil together with metformin improved pulmonary artery acceleration time to ejection time ratio and tricuspid annular plane systolic excursion with AMPK (AMP-activated protein kinase) activation in skeletal muscle and the right ventricle. CONCLUSIONS: Our data suggest a potential use of treprostinil as an early treatment for mild metabolic syndrome-associated PH-HFpEF and that combined treatment with treprostinil and metformin may improve hyperglycemia and cardiac function in a more severe disease.

Quantification of Coupled Stiffness and Fiber Orientation Remodeling in Hypertensive Rat Right-Ventricular Myocardium Using 3D Ultrasound Speckle Tracking with Biaxial Testing.

Mechanical and structural changes of right ventricular (RV) in response to pulmonary hypertension (PH) are inadequately understood. While current standard biaxial testing provides information on the mechanical behavior of RV tissues using surface markers, it is unable to fully assess structural and mechanical properties across the full tissue thickness. In this study, the mechanical and structural properties of normotensive and pulmonary hypertension right ventricular (PHRV) myocardium through its full thickness were examined using mechanical testing combined with 3D ultrasound speckle tracking (3D-UST). RV pressure overload was induced in Sprague-Dawley rats by pulmonary artery (PA) banding. The second Piola-Kirchhoff stress tensors and Green-Lagrangian strain tensors were computed in the RV myocardium using the biaxial testing combined with 3D-UST. A previously established non-linear curve-fitting algorithm was applied to fit experimental data to a Strain Energy Function (SEF) for computation of myofiber orientation. The fiber orientations obtained by the biaxial testing with 3D-UST compared well with the fiber orientations computed from the histology. In addition, the re-orientation of myofiber in the right ventricular free wall (RVFW) along longitudinal direction (apex-to-outflow-tract direction) was noticeable in response to PH. For normotensive RVFW samples, the average fiber orientation angles obtained by 3D-UST with biaxial test spiraled from 20° at the endo-cardium to -42° at the epi-cardium (Δ = 62°). For PHRV samples, the average fiber orientation angles obtained by 3D-UST with biaxial test had much less spiral across tissue thickness: 3° at endo-cardium to -7° at epi-cardium (Δ = 10°, P<0.005 compared to normotensive).