An Effective Deep Neural Network for Lung Lesions Segmentation From COVID-19 CT Images.

Automatic segmentation of lung lesions from COVID-19 computed tomography (CT) images can help to establish a quantitative model for diagnosis and treatment. For this reason, this article provides a new segmentation method to meet the needs of CT images processing under COVID-19 epidemic. The main steps are as follows: First, the proposed region of interest extraction implements patch mechanism strategy to satisfy the applicability of 3-D network and remove irrelevant background. Second, 3-D network is established to extract spatial features, where 3-D attention model promotes network to enhance target area. Then, to improve the convergence of network, a combination loss function is introduced to lead gradient optimization and training direction. Finally, data augmentation and conditional random field are applied to realize data resampling and binary segmentation. This method was assessed with some comparative experiment. By comparison, the proposed method reached the highest performance. Therefore, it has potential clinical applications.

Fabrication of Three-Dimensional Flower-like Heterogeneous Fe3O4/Fe Particles with Tunable Chemical Composition and Microwave Absorption Performance.

Heterogeneous Fe3O4 and Fe composites are highly desirable for microwave absorption application because of their complementary electromagnetic (EM) properties. With three-dimensional (3D) Fe2O3 as a sacrificing template, we realize the construction of Fe3O4/Fe composites with tunable chemical composition, and more importantly, these composites inherit the unique 3D microstructure from their precursor. The change in chemical composition produces significant impacts on the EM functions of these composites. On the one hand, dielectric loss can be improved greatly through positive interfacial polarization and reach the peak when the mass contents of Fe3O4 and Fe are 72.1 and 27.9 wt %, respectively. On the other hand, high Fe content slightly pulls down magnetic loss in the low-frequency range but favors strong magnetic loss in the high-frequency range because of the breakthrough of Snoek's limitation. The attenuation constant reveals that dielectric loss dominates overall consumption of incident EM waves. As a result, the optimized composite, F-350 (the reduction of Fe2O3 is conducted at 350 °C), shows the best microwave absorption performance, whose strongest reflection loss is -56.0 dB at 17.5 GHz and the effective bandwidth can cover the frequency range of 12.0-15.5 GHz with the thickness of 1.5 mm. Furthermore, an ultrawide effective bandwidth of 15.3 GHz can be achieved with the integrated thickness of 1.0-5.0 mm. Such a performance is superior to those of many reported Fe3O4/Fe composites, and a comparative analysis manifests that good microwave absorption of F-350 is also benefited from its unique 3D architecture.