Gaussian sliding window for robust processing laser speckle contrast images

Abstract

The laser speckle contrast analysis (LASCA) is one of the most applicable tools in microcirculation studies. While the basic idea, as well as experimental setup for this method, are fairly simple, there is still the room for advancing of data processing algorithms. Specifically, the conventional realizations of LASCA method may limit the spatial and/or temporal resolution and thus fail in the detection of very small contrast objects since they based on the fixed‐size rectangular sliding window function. We suggest an alternative data processing algorithm based on the usage of the Gaussian sliding filter for a sequential determination of both spatial and temporal parts of the speckle contrast. The suggested replacement of conventional box filter leads to the monotonic damping of high‐frequency spectral components that results in a better elimination of ringing and aliasing effects in the spatio‐temporal speckle contrast outputs. Additionally, we show that such sliding filtration increases robustness with respect to the processing of a sequence of nonstabilised images. We support this consideration with representative examples of processing both surrogate and real experimental data.

Untitled

“…This limitation is due to the utilization of local windows for spatial averaging in stLASCA, which enhances the background signal and leads to a higher value approximating the blood flow signal. The method that uses a Fourier-based Gaussian sliding window to calculate the contrast can partially solve the above problem [15] . However, this method can make the processed image look blurry and distorted because it can efficiently smooth the noise.…”

“…On the other hand, unlike the 1D-FFT used in LSCI based on intensity fluctuation modulation (LSCI-IFM) and 2D-FFT used in Ref. [15], the 3D-FFT used in LSCI-uSTFT can reduce the computation complexity. Moreover, since the FFT is linear, the combination of spatial and temporal Fourier transform prolongs the exposure time of the speckle images in a computational sense while avoiding the issues of overexposure, saturation, and reduced contrast that may arise from physical overexposure.…”

“…In conventional LSCI techniques, the laser speckle spatial contrast analysis (LSSCA) [12] , laser speckle temporal contrast analysis (LSTCA) [13] , and spatiotemporal laser speckle contrast analysis (stLASCA) [14] all fall into the domain of filtration procedures that use the rectangular window to calculate the contrast that cannot achieve high-quality imaging, high statistical accuracy, and fast imaging. Recently, the Fourier-based Gaussian sliding window [15] and the adaptive window [16] were introduced into the calculation of LSCI, which has made a great breakthrough in developing LSCI technology. On the other hand, statistical accuracy, imaging quality, and reconstruction speed are closely related to the frame number of the raw speckle images [17][18][19] .…”

Supplementary document for Real-time monitoring of biomechanical activity in aphids by laser speckle contrast imaging - 5356065.pdf

“…Another way to further improve the contrast accuracy and reduce noise in sLASCA/tLASCA/stLASCA is to apply a Gaussian shape sliding window [32]. The proposed alternative sliding window can suppress high-frequency spectral noise components, ringing, and aliasing.…”

“…The proposed alternative sliding window can suppress high-frequency spectral noise components, ringing, and aliasing. In combination with the fact that Gaussian filtering improves the spatial and/or temporal localization resolution of an object compared to rectangular filters, this leads to the detection of extra-fine contrast structures [32].…”