We report on the performance of a vector apodizing phase plate coronagraph that operates over a wavelength range of 2-5 μmand is installed in MagAO/Clio2 at the 6.5 m Magellan Clay telescope at Las Campanas Observatory, Chile. The coronagraph manipulates the phase in the pupil to produce three beams yielding two coronagraphic point-spread functions (PSFs) and one faint leakage PSF. The phase pattern is imposed through the inherently achromatic geometric phase, enabled by liquid crystal technology and polarization techniques. The coronagraphic optic is manufactured using a direct-write technique for precise control of the liquid crystal patternand multitwist retarders for achromatization. By integrating a linear phase ramp to the coronagraphic phase pattern, two separated coronagraphic PSFs are created with a single pupil-plane optic, which makes it robust and easy to install in existing telescopes. The two coronagraphic PSFs contain a 180°dark hole on each side of a star, and these complementary copies of the star are used to correct the seeing halo close to the star. To characterize the coronagraph, we collected a data set of a bright (m L =0-1) nearby star with ∼1.5 hr of observing time. By rotating and optimally scaling one PSFand subtracting it from the other PSF, we see a contrast improvement by 1.46 magnitudes at l D 3.5. With regular angular differential imaging at 3.9 μm, the MagAO vector apodizing phase plate coronagraph delivers a s D 5 mag contrast of 8.3 (= -10 3.3 ) at 2 l Dand 12.2 (= -10 4.8 ) at l D 3.5.
With traditional beamforming methods, ultrasound B-mode images contain speckle noise caused by the random interference of subresolution scatterers. In this paper, we present a framework for using neural networks to beamform ultrasound channel signals into speckle-reduced B-mode images. We introduce log-domain normalization-independent loss functions that are appropriate for ultrasound imaging. A fully convolutional neural network was trained with simulated channel signals that were co-registered spatially to ground truth maps of echogenicity. Networks were designed to accept 16 beamformed subaperture radiofrequency signals. Training performance was compared as a function of training objective, network depth, and network width. The networks were then evaluated on simulation, phantom, and in vivo data and compared against existing speckle reduction techniques. The most effective configuration was found to be the deepest (16 layer) and widest (32 filter) networks, trained to minimize a normalization-independent mixture of the ℓ 1 and multi-scale structural similarity losses. The neural network significantly outperformed delay-and-sum and receive-only spatial compounding in speckle reduction while preserving resolution and exhibited improved detail preservation over a non-local means methods. This work demonstrates that ultrasound B-mode image reconstruction using machine-learned neural networks is feasible and establishes that networks trained solely in silico can be generalized to real-world imaging in vivo to produce images with significantly reduced speckle.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.