Real-time GPU-accelerated processing and volumetric display for wide-field laser-scanning optical-resolution photoacoustic microscopy

Abstract

Fast signal processing and real-time displays are essential for practical imaging modality in various fields of applications. However, the imaging speed in optical-resolution photoacoustic microscopy (OR-PAM), in particular, depends on factors such as the pulse repetition rate of the laser, scanning method, field of view (FOV), and signal processing time. In the past, efforts to increase acquisition speed either focused on developing new scanning methods or using lasers with higher pulse repetition rates. However, high-speed signal processing is also important for real-time volumetric display in OR-PAM. In this study, we carried out parallel signal processing using a graphics processing unit (GPU) to enable fast signal processing and wide-field real-time displays in laser-scanning OR-PAM. The average total GPU processing time for a B-mode PAM image was approximately 1.35 ms at a display speed of 480 fps when the data samples were acquired with 736 (axial) × 500 (lateral) points/B-mode-frame at a pulse repetition rate of 300 kHz. In addition, we successfully displayed maximum amplitude projection images of a mouse's ear as volumetric images with an FOV of 3 mm × 3 mm (500 × 500 pixels) at 1.02 s, corresponding to 0.98 fps.

A continuous adjoint for photo-acoustic tomography of the brain

“…In this work, we derived the adjoint of the continuous map defined in (16) and (17), which describes the propagation of PA waves in linear isotropic viscoelastic media with the absorption and physical dispersion following a frequency power law. We analytically showed that a numerical computation of our continuous adjoint using a k-space pseudo-spectral method matches the algebraic adjoint of an associated discretised map defined by (63) and (65).…”

“…In addition, in practical cases, to extract all information available from the measured data, the maximal frequency supported by the computational grid must match the maximal frequency that is detectable by detectors [33]. This dramatically increases the computational demands regarding storage space and speed, but it can be handled using GPU accelerated computing [16], or Field-programmable gate array (FPGA) [34]. The 3D detection setting in our study simulated a planar Fabry-Pérot (FP) photoacoustic scanner, which requires several minutes to collect time series of data from PA wavefields [1].…”

Autofocus method for automated microscopy using embedded GPUs

“…Dogar et al [9], improved that result up to 70frps (1024x1024 resolution) employing a 1334-core GPU. Similarly, Kang et al [13] employed a GPU populated with 2880 cores for real-time volumetric display of optical-resolution photoacoustic microscopy (OR-MAP). They reached an impressive refreshing rate of 480fps of low resolution (736x500) images.…”

Parallel Computing for Quantitative Blood Flow Imaging in Photoacoustic Microscopy

“…Currently, several parallel computing methods with GPU have been implemented to reconstruct images in PAT systems such as back-projection (BP)-based PAT [19], finite element method (FEM)-based time-domain quantitative PAT [21] and double-state delay-multiply-and-sum (DS-DMAS)-based PAT [23]. In PAM system, GPUs are mainly adopted for the real-time structure imaging such as displaying maximum amplitude projection (MAP) images of blood vessels in a mouse's ear [24,25]. High performance computation of quantitative blood flow imaging in PAM has not been reported and remains a challenge.In this work, we propose a GPU-based parallel computing design for quantitative blood flow imaging in PAM.…”

“…The multiplied signal is then converted back to the time domain via the inverse-FFT [24]. The total number of A-line is marked as N. For this specific example, N equals 7.2 × 10 5 .…”