Experimental characterization of a shape optimized acoustic lens: Application to compact speakerphone design.

This work presents the shape optimization and subsequent experimental validation of an acoustic lens with application to a compact loudspeaker, such as found in commercial speakerphones. The shape optimization framework is based on a combined lumped parameter and boundary element method model using free form deformation geometry parameterization. To test the optimized design, the loudspeaker lens is three-dimensionally printed and experimentally characterized under anechoic conditions on a finite baffle with respect to its off-axis frequency response. The overall tendencies of the frequency responses agree well between measurement and simulations within the optimization frequency range and at low frequencies. The optimization process is applied to a model including acoustic lumped parameter approximations. The shortcomings of the assumptions made in the model are revealed by laser Doppler vibrometer measurements of the loudspeaker driver and modelling of the mechanical vibrations of the lens.

Deep learning based cough detection camera using enhanced features.

Coughing is a typical symptom of COVID-19. To detect and localize coughing sounds remotely, a convolutional neural network (CNN) based deep learning model was developed in this work and integrated with a sound camera for the visualization of the cough sounds. The cough detection model is a binary classifier of which the input is a two second acoustic feature and the output is one of two inferences (Cough or Others). Data augmentation was performed on the collected audio files to alleviate class imbalance and reflect various background noises in practical environments. For effective featuring of the cough sound, conventional features such as spectrograms, mel-scaled spectrograms, and mel-frequency cepstral coefficients (MFCC) were reinforced by utilizing their velocity (V) and acceleration (A) maps in this work. VGGNet, GoogLeNet, and ResNet were simplified to binary classifiers, and were named V-net, G-net, and R-net, respectively. To find the best combination of features and networks, training was performed for a total of 39 cases and the performance was confirmed using the test F1 score. Finally, a test F1 score of 91.9% (test accuracy of 97.2%) was achieved from G-net with the MFCC-V-A feature (named Spectroflow), an acoustic feature effective for use in cough detection. The trained cough detection model was integrated with a sound camera (i.e., one that visualizes sound sources using a beamforming microphone array). In a pilot test, the cough detection camera detected coughing sounds with an F1 score of 90.0% (accuracy of 96.0%), and the cough location in the camera image was tracked in real time.

Surface Design of Eu-Doped Iron Oxide Nanoparticles for Tuning the Magnetic Relaxivity.

Relaxivity tuning of nanomaterials with the intrinsic T1- T2 dual-contrast ability has great potential for MRI applications. Until now, the relaxivity tuning of T1 and T2 dual-modal MRI nanoprobes has been accomplished through the dopant, size, and morphology of the nanoprobes, leaving room for bioapplications. However, a surface engineering method for the relaxivity tuning was seldom reported. Here, we report the novel relaxivity tuning method based on the surface engineering of dual-mode T1- T2 MRI nanoprobes (DMNPs), along with protein interaction monitoring with the DMNPs as a potential biosensor application. Core nanoparticles (NPs) of europium-doped iron oxide (EuIO) are prepared by a thermal decomposition method. As surface materials, citrate (Cit), alendronate (Ale), and poly(maleic anhydride- alt-1-octadecene)/poly(ethylene glycol) (PP) are employed for the relaxivity tuning of the NPs based on surface engineering, resulting in EuIO-Cit, EuIO-Ale, and EuIO-PP, respectively. The key achievement of the current study is that the surface materials of the DMNP have significant impacts on the r1 and r2 relaxivities. The correlation between the hydrophobicity of the surface material and longitudinal relaxivity ( r1) of EuIO NPs presents an exponential decay feature. The r1 relaxivity of EuIO-Cit is 13.2-fold higher than that of EuIO-PP. EuIO can act as T1- T2 dual-modal (EuIO-Cit) or T2-dominated MRI contrast agents (EuIO-PP) depending on the surface engineering. The feasibility of using the resulting nanosystem as a sensor for environmental changes, such as albumin interaction, was also explored. The albumin interaction on the DMNP shows both T1 and T2 relaxation time changes as mutually confirmative information. The relaxivity tuning approach based on the surface engineering may provide an insightful strategy for bioapplications of DMNPs and give a fresh impetus for the development of novel stimuli-responsive MRI nanoplatforms with T1 and T2 dual-modality for various biomedical applications.

Survival of Verwey transition in gadolinium-doped ultrasmall magnetite nanoparticles.

We have demonstrated that the Verwey transition, which is highly sensitive to impurities, survives in anisotropic Gd-doped magnetite nanoparticles. Transmission electron microscopy analysis shows that the nanoparticles are uniformly distributed. X-ray photoelectron spectroscopy and EDS mapping analysis confirm Gd-doping on the nanoparticles. The Verwey transition of the Gd-doped magnetite nanoparticles is robust and the temperature dependence of the magnetic moment (zero field cooling and field cooling) shows the same behaviour as that of the Verwey transition in bulk magnetite, at a lower transition temperature (∼110 K). In addition, irregularly shaped nanoparticles do not show the Verwey transition whereas square-shaped nanoparticles show the transition. Mössbauer spectral analysis shows that the slope of the magnetic hyperfine field and the electric quadrupole splitting change at the same temperature, meaning that the Verwey transition occurs at ∼110 K. These results would provide new insights into understanding the Verwey transition in nano-sized materials.

Chaotic indices and canonical ensemble of heart rate patterns in small-for-gestational age fetuses.

AIMS: We studied how linear and nonlinear heart rate dynamics differ between normal fetuses (n=135) and uncomplicated small-for-gestational age (SGA) fetuses (n=65), aged 32-40 weeks' gestation. METHODS: We analyzed each fetal heart rate time series for 20 min. We quantified the complexity (nonlinear dynamics) of each fetal heart rate (FHR) time series by approximate entropy (ApEn) and correlation dimension (CD). The linear dynamics were analyzed by canonical correlation analysis (CCA). RESULTS: The ApEn and CD of the uncomplicated SGA fetuses were significantly lower than that of the normal fetuses in all three gestational periods (32-34, 35-37, 38-40 weeks). Canonical correlation ensemble in SGA fetuses is slightly higher than normal ones in all three gestational periods, especially at 35-37 weeks. CONCLUSIONS: Irregularity and complexity of the heart rate dynamics of SGA fetuses are lower than that of normal ones. Also, canonical ensemble in SGA fetuses is higher than in normal ones, suggesting that the FHR control system has multiple complex interactions. Along with the clear difference between the two groups' non-linear chaotic dynamics in FHR patterns, we clarified the hidden subtle differences in linearity (e.g., canonical ensemble). The decrease in non-linear dynamics may contribute to the increase in linear dynamics. The present statistical methodology can be readily and routinely utilized in obstetrics and gynecologic fields.