Hanbing Chu

the application of medical ultrasound imaging instruments, microvessel detection is the basis for accurate measurement of vessel parameters and hemodynamic parameters, which directly affects the instrument's ability to diagnose diseases early and evaluate treatment effects. Ultrasound localization microscopy enables superresolution imaging and detecting of deep microvessels via microbubble (MB) localization and tracking, but with low temporal resolution. To balance spatial and temporal resolution, higher-order statistics improves spatial resolution and enables the reconstruction of a microvessel image with shorter acquisitions. However, in plane wave imaging, the point spread function (PSF) broadens and changes with depth, limiting spatial resolution gains. To overcome this and further improve spatiotemporal resolution, a nonlocalized dynamic microvessel imaging method called MB gradient convergence and accumulation reconstruction (GCAR) was proposed. The GCAR method is implemented via spatial and temporal transformations of data at high MB concentrations; the spatial transformation anisotropically shrinks and separates MBs toward the center by using the symmetry of the MB gradient, and the temporal transformation reconstructs microvessels by measuring the statistical independence of the MB signals through high-order cross accumulation. The simulation results revealed that GCAR decreased the full width at half maximum of the MBs by 80%, and was able to separate parallel vessels with spacings of 0.25λ (4 MHz) and 0.125λ (10 MHz). The in vivo results revealed that GCAR could quickly reconstruct the microvasculature with a spatial resolution of 18.19 μm (12 MHz) for the rat brain and 76.28 μm (4 MHz) for the rabbit kidney. GCAR dynamically displays blood flow with a temporal resolution of 50 ms. The GCAR approach improves the spatiotemporal resolution of medical ultrasound imaging instruments, detects smaller and more microvessels with shorter acquisition times, which may provide a useful tool for hemodynamic measurement and examination of microvascular diseases.

Cerebral microvasculature imaging is crucial in the diagnosis of various cerebrovascular diseases. Ultrasound localization microscopy (ULM) enables microvascular imaging at microscopic resolution through the localization of injected microbubbles. However, the strong attenuation of transcranial ultrasound limits the sensitivity of vascular detection, and phase aberration caused by mismatched speed of sound in different tissues reduces the resolution of transcranial ULM (TULM). In this study, a chirp-coded excitation signal was transmitted, and an adaptive attenuation mismatched filter (AMF) was constructed based on the attenuation of each frequency component to improve the signal-to-noise ratio (SNR) and resolution of the compressed signal. Moreover, a multilayer correlation phase correction method was proposed for precision TULM, independently correcting phase aberration in different fields of view. The simulation results confirm that the chirp-coded excitation and AMF increase the SNR by 11.31 dB and the axial resolution by 0.08λ compared with those of the single-pulse excitation under strong attenuation conditions. For phase aberration, the multilayer correlation phase correction improves the lateral and axial resolutions to 0.77λ and 0.19λ, respectively. Additionally, the advanced phase correction method enhances the refocusing ability, improving the contrast by 2.46 dB. For in vivo TULM, the proposed method was more robust, that can reconstruct richer microvascular structures than the conventional method, with a 15.68% increase in saturation. Moreover, the resolution of the microvasculature improved by 7.19 μm (27.24%). These results validate the advantages of the proposed method in achieving rich microvascular reconstruction with higher resolution than conventional methods under transcranial imaging conditions. Thus, the proposed method has application potential in ultrasound imaging of small vessels in the brain.

Objective. Estimating the high-resolution (HR) blood flow velocity and pressure fields for the diagnosis and treatment of vascular diseases remains challenging. Approach. In this study, a physics-informed neural network (PINN) with a refined mapping capability was combined with ultrafast ultrasound image velocimetry (u-UIV) to predict HR hemodynamic parameters. Specifically, the Navier–Stokes equations were encoded into the PINN to dynamically optimize the network performance under physical constraints, and a refined mapping network was added at the input to achieve data refinement. During the prediction of HR ultrasound hemodynamic parameters, only the sparse spatial coordinates in the time series were input into the PINN, and the velocity vectors generated from the u-UIV were used together with physical residuals to enhance the physical correctness of HR predictions during the iterative process. Main results. The performance of the refined mapping network was validated via simulations, with a 1.9-fold increase in the radial resolution and a 2.5-fold increase in the axial resolution. HR velocity field estimation from in vitro and in vivo data showed good agreement with theoretical values and u-UIV measurements, with micrometer-level spatial resolution (88 µm × 115 µm for straight vessel, 75 µm × 120 µm for stenotic vessel and 63 µm × 79 µm for in vivo data), while the pressure field could be inferred from physical laws. Significance. The proposed method performs well when few data samples are available and has the potential to assist in the clinical diagnosis of vascular diseases.

Simultaneously improving the spatial and temporal resolution of ultrafast power Doppler imaging (UPDI) is conducive to monitoring rapid changes in microvascular morphology and dynamics and increasing quantification accuracy. Grouped angle singular value decomposition (GA-SVD) has been studied to suppress clutter at enhanced temporal resolution, but it struggles to remove spatially varying noise. This study presents an adaptive weight-based nonlinear compounding method called generalized frame-multiply-and-sum (FMAS) factor based on pth root compression (p-GFMAS) to address the above legacy noise and improve the spatial resolution of UPDI at better temporal resolution. p-GFMAS utilizes angular coherence extracted based on the autocorrelation function to emphasize coherent blood signals and suppress incoherent noise, where the adaptive effect is adjusted by p-value. The effectiveness of the p-GFMAS in improving spatial resolution was verified through simulations. And the superiority of the p-GFMAS united with GA-SVD in enhancing spatiotemporal resolution was validated with in vivo studies. Simulation results demonstrated that p-GFMAS outperforms other compounding methods, and p-GFMAS increased the signal-to-noise ratio (SNR) by 22.77 dB (234%) and reduced the full-width at half-maximum (FWHM) by $274~\mu $ m (43.4%) over traditional coherent compounding (CC). Rat brain and human liver results depicted that p-GFMAS united with GA-SVD achieved highest image quality when using shorter ensembles due to the increased SNR and reduced FWHM. Dynamic results further revealed the ability of the united method to capture flow-dynamics of the rat brain at a frame rate of 20 Hz. Overall, p-GFMAS united with GA-SVD realizes high spatiotemporal resolution contrast-free microvascular visualization, facilitating an understanding of the mechanisms of various physiological and pathological processes.