Search

In recent years, computer-aided automatic polyp segmentation and neoplasm detection have been an emerging topic in medical image analysis, providing valuable support to colonoscopy procedures. Attentions have been paid to improving the accuracy of polyp detection and segmentation. However, not much focus has been given to latency and throughput for performing these tasks on dedicated devices, which can be crucial for practical applications. This paper introduces a novel deep neural network architecture called BlazeNeo, for the task of polyp segmentation and neoplasm detection with an emphasis on compactness and speed while maintaining high accuracy. The model leverages the highly efficient HarDNet backbone alongside lightweight Receptive Field Blocks and a feature aggregation mechanism for computational efficiency. An auxiliary training strategy is proposed to take full advantage of the training data for the segmentation quality. Our experiments on a challenging dataset show that BlazeNeo achieves improvements in latency and model size while maintaining comparable accuracy against state-of-the-art methods. We obtain over 155 fps while outperforming all compared models in terms of accuracy in INT8 precision when deploying on a dedicated edge device with a conventional configuration.

The modified Arrott plot method was used to analyze magnetic-field dependences of magnetization, \(M\) ( \(H\) ), for two polycrystalline samples Pr1-x Srx CoO3 with \(x = 0.4\) and 0.5. The analyses determined the values of critical parameters \(T_{C} \approx 204\) K, \(\beta = 0.654~\pm ~0.016\) , \(\gamma = 1.11~\pm ~0.009\) , and \(\delta = 2.70~\pm ~0.009\) for \(x = 0.4\) , and \(T_{C} \approx 218\) K, \(\beta = 0.495~\pm ~0.004\) , \(\gamma =0.991~\pm ~0.003\) , and \(\delta = 2.86~\pm ~0.009\) for \(x = 0.5\) . With these critical values, the \(M(H\) ) data around the \(T_{C}\) of the samples fall into two universal branches of a scaling function \(M(H\) , \(\varepsilon ) ={} \vert \varepsilon \vert ^{\beta }\!f_{\pm }\) ( \(H/ \vert \varepsilon \vert ^{\beta +\gamma }\) ), where \(\varepsilon =(T\) – \(T_{C}\) )/ \(T_{C}\) , and \(f_{+}\) for \(T > T_{C}\) and \(f_{-}\) for \(T < T_{C}\) . Here, the \(T_{C}\) increase with the increasing Sr content is ascribed to an increase of ferromagnetic (FM) double-exchange interaction pairs of Co3+–Co4+, and to the \(\beta \) change from 0.654 (for \(x = 0.4\) ) to 0.495 (for \(x = 0.5\) ). This corresponds to the change in FM interactions from short-range to nearly long-range FM orders. Based on the \(M(H)\) data, we also determined the magnetic-entropy change, which reaches a maximum value ( \(\vert \Delta S_{\max } \vert \) ) around the \(T_{C}\) ; \(\vert \Delta S_{\max } \vert \,\approx ~1.9\) and 2.2 J/kg \(\cdot \) K for \(x =0.4\) and 0.5, respectively, for a magnetic field change \(\Delta H = 50\) kOe. The relative cooling power thus increases from 52 J/kg (for \(x = 0.4\) ) to 84 J/kg (for \(x = 0.5\) ). In particular, field dependences of the magnetic-entropy change ( \(\Delta S_{m}\) ) obey a power function \(\vert\Delta S_{m}\) ( \(H\) ) \(\vert ~\propto H^{\mathbf {n}}\) , where the magnetic-ordering parameter ( \(n\) ) slightly decreases from 0.75 for \(x = 0.4\) to 0.70 for \(x = 0.5\) . [ABSTRACT FROM AUTHOR]

This paper points out that the magnetic-phase transition and magnetocaloric effect of La0.7Ca0.3MnO3 (LCMO) can be easily controlled by using the mechanical milling method. Changing the milling time from 5 to 30 min, we have obtained LCMO nanoparticles (NPs) with average crystallite sizes (d, determined by the Williamson–Hall method) ranging from 100 to 45 nm. The magnetic studies (based on a superconducting quantum interference device) and simple analyses (based on Banerjee’s criteria) prove the magnetic-phase transformation from the first-order to the second-order, which takes place at a threshold value of d located in the range 60−70 nm. Compared with the as-prepared LCMO sample (a first-order magnetic phase transition), though the d decrease reduces the values of the T \(_{\mathrm {\textit{C}}}\) , magnetization, magnetic-entropy change, and refrigerant capacity, but the width of the magnetic phase transition is increased remarkably. This widens the working range of LCMO NPs in magnetic refrigeration applications. We believe that the presence of surface-related effects, lattice strain, and distortions leads to Mn3+–Mn4+ ferromagnetic interactions in LCMO NPs weaker than that in the as-prepared sample. [ABSTRACT FROM AUTHOR]

We report here on a new approach for the detection and identification of actinides (^239Pu, ^241Am, ^244Cm, etc). This approach is based on the use of a novel device consisting of a boron doped nanocrystalline diamond film deposited onto a silicon PIN diode alpha particle sensor. The actinides concentration is probed in situ in the measuring solution using a method based on electro-precipitation that can be carried out via the use of a doped diamond electrode. The device allows probing directly both alpha-particles activity and energy in liquid solutions. In this work, we address the optimization of the actinides electro-precipitation step onto the sensor. The approach is based on fine tuning the pH of the electrolyte, the nature of the supporting electrolytes (Na2SO4 or NaNO3), the electrochemical cell geometry, the current density value, the precipitation duration as well as the sensor surface area. The deposition efficiency was significantly improved with values reaching for instance up to 81.5% in the case of electro-precipitation of 5.96 Bq^241Am on the sensor. The diamond/silicon sensor can be reused after measurement by performing a fast decontamination step at high yields \geq 99%, where the ^241Am electro-precipitated layer is quickly removed by applying an anodic current (+2 mA · cm^-2 for 10 minutes) to the boron doped nanocrystalline diamond electrode in aqueous solution. This study demonstrated that alpha-particle spectroscopic measurements could be made feasible for the first time in aqueous solutions after an electrochemical deposition process, with theoretical detections thresholds as low as 0.24 Bq · L^-1. We believe that this approach can be of very high interest for alpha-particle spectroscopy in liquids for actinides trace detection. [ABSTRACT FROM PUBLISHER]

A technique called weak-lock (WL) has been proposed and applied to a simple optical on/off switch to increase its working speed from 1 to 12 Gb/s with clear output waveforms, whereas the required injection power is decreased. The switch is based on the injection-locking property of a single-mode Fabry-Pérot laser diode (SMFP-LD). Unlike the case of full-lock (FL) in which self-locked mode and all side modes of the SMFP-LD are fully suppressed due to high injected input power, in the WL case, the injection power is gradually decreased as if the SMFP-LD's self-locked mode is still maintained. It was observed that the output waveform's rising edge, taken at the self-locked mode's wavelength after the filter, in the WL case is clean at speed of 12 Gb/s compared to the blur and noisy rising edge in the FL case even at much lower speed (1 Gb/s). In summary, the WL technique shows advantages compared to the conventional FL technique in terms of lower injection power and higher on/off switching speed. [ABSTRACT FROM PUBLISHER]