Optics Communications 439 (2019) 38–46
Contents lists available at ScienceDirect
Optics Communications
journal homepage: www.elsevier.com/locate/optcom
Laser speckle imaging for blood flow based on pixel resolved zero-padding
auto-correlation coefficient distribution
Ru Zhang
a
, Lipei Song
a,c,
∗
, Jiachun Xu
b
, Xiaoying An
a
, Weiming Sun
b
, Xing Zhao
a
,
Zhen Zhou
b,
∗∗
, Lingling Chen
d
a
Institute of Modern Optics, Nankai University, Tianjin, 300350, China
b
Department of Acupuncture Cerebropathy, Second Affiliated Hospital of Tianjin University of TCM, Tianjin, 300150, China
c
State Key Laboratory of Quantum Optics and Quantum Optics Devices, Institute of Opto-Electronics, Shanxi University, Taiyuan, 030006, China
d
College of Health Science and Environmental Engineering, Shenzhen Technology University, Shenzhen, 518118, China
A R T I C L E I N F O
Keywords:
Laser speckle imaging
Auto-correlation imaging
Zero padding
A B S T R A C T
In this paper we present a method of zero-padding temporal auto-correlation imaging (ztDCI) for imaging blood
flows. Compared with temporal auto-correlation imaging (tDCI), ztDCI effectively slows down the decorrelation
speed of speckle frames, so it has wider applicable speed range. We also prove that the combination of tDCI
and ztDCI can compensate the non-monotonous trend of ztDCI VS speed and provides higher relative signal to
noise rate (rSNR) for slow flows and short exposure times than laser speckle contrast imaging (LSCI). It captures
both slow and fast flows independent of camera exposure time which will be an advantage for in vivo or clinical
applications.
1. Introduction
Imaging blood vessels and detecting blood flows has attracted in-
tense interest. Among a variety of methods such as Doppler imaging [1,
2] and OCT [3–5], laser speckle contrast imaging (LSCI) is a full-field
and economical technique. There are different ways to analyze laser
speckle images for imaging blood vessels depending on the purpose. For
instance, spatial laser speckle contrast imaging (sLSCI) [6,7] has higher
temporal resolution because it is based on a single speckle image, but
has poor spatial resolution limited by the kernel size in the calculation.
Temporal laser speckle contrast imaging (tLSCI) [8], on the contrary,
offers pixel resolved images but sacrifices the temporal resolution. In
LSCI, the exposure time of the camera highly affects the vessels that
can be imaged [9,10]. To resolve small vessels or slow flows, a long
exposure time is necessary; however this induces a loss of temporal
resolution and sensitivity to fast flows. In addition, the optimal exposure
time for different flow speeds is difficult to pre-determine. Laser speckle
imaging based on intensity fluctuations has higher spatial resolution and
SNR [11], but it usually needs thousands of speckle images, and the
requirement for a Fourier transform and frequency separation increases
the processing complexity.
In this paper, we provide a method (temporal digital correlation
imaging, termed tDCI) to image the blood vessels from laser speckle
images based on calculating the temporal auto-correlation coefficient
in an imaging domain. This approach is different from digital image
∗
Corresponding author at: Institute of Modern Optics, Nankai University, Tianjin, 300350, China.
∗∗
Corresponding author.
E-mail addresses: gm_imo@nankai.edu.cn (L. Song), zhouzhen7681@126.com (Z. Zhou).
correlation (DIC) that is to calculate the correlation coefficient of
subset kernels [12,13]. tDCI is a pixelwise calculation from temporal
sequences of speckle images, on the assumption that the pixel intensities
of the temporal speckle sequence are weak (or wide)-sense stationary
(WSS), meaning that the mean and the auto-variance do not vary with
respect to time and the auto-correlation only depends on the time
elapsed. This is reasonable when there is no a sudden environmental
disturbance. Therefore tDCI has pixel-level spatial resolution. However,
the ordinary temporal auto-correlation coefficient decreases rapidly to
zero or a stable level [14] depending on factors such as the flow speed
or stationary contributions. This leads to a low dynamic range for tDCI
when investigating blood circulation.
To solve this problem, we introduce a zero-padding to each of the
intensity arrays in the correlation calculation, which we call ztDCI. Zero-
padding does not introduce extra information to the intensity array and
is easy to analyze. Combining tDCI and ztDCI, small vessels and slow
flows can be detected independent to exposure times. We find that a
disadvantage of ztDCI is its non-monotonous trend with flow speeds, but
combining tDCI with ztDCI can both effectively compensate this non-
monotonous trend and maintain the large dynamic range of ztDCI. We
compare this method with tLSCI and prove with theoretical derivation
and experimental results that more small vessels can be detected and
both the relative signal to noise rate and the vessel-tissue contrast are
also improved with short exposure times.
https://doi.org/10.1016/j.optcom.2019.01.037
Received 31 October 2018; Received in revised form 10 January 2019; Accepted 13 January 2019
Available online 18 January 2019
0030-4018/© 2019 Elsevier B.V. All rights reserved.