Open Access
Add to BibSonomy
Abstract
Laser speckle contrast imaging (LSCI) has gained significant attention in the biomedical field for its ability to map the spatio-temporal dynamics of blood perfusion in vivo. However, LSCI faces difficulties in accurately resolving blood perfusion in microvessels. Although the transmissive detecting geometry can improve the spatial resolution of tissue imaging, ballistic photons directly transmitting forward through tissue without scattering will cause misestimating in the flow speed by LSCI because of the lack of a quantitative theoretical model of transmissvie LSCI. Here, we develop a model of temporal LSCI which accounts for the effect of nonscattered light on estimating decorrelation time. Based on this model, we further propose a dual-exposure temporal laser speckle imaging method (dEtLSCI) to correct the overestimation of background speed when performing traditional transmissive LSCI, and reconstruct microvascular angiography using the scattered component extracted from total transmitted light. Experimental results demonstrated that our new method opens an opportunity for LSCI to simultaneously resolve the blood vessels morphology and blood flow speed at microvascular level in various contexts, ranging from the drug-induced vascular response to angiogenesis and the blood perfusion monitoring during tumor growth.
© 2024 Optica Publishing Group under the terms of the Optica Open Access Publishing Agreement
1. Introduction
With the advantages of rapid imaging and simplicity of wide-field recording, laser speckle contrast imaging (LSCI) recently attracts more and more applications in biomedical research and theranostics [1–5]. In LSCI, the decorrelation time of electric field of scattered light is estimated through spatio-temporal statistics of time-varying speckles to resolve the speed of moving particles in biological tissues [6,7]. Nevertheless, the inherent dominance of backscattered photons from superficial tissue layers in wide-field reflective geometry compromises the depth capability of reflective LSCI, constraining its probing depth to the order of a few hundred micrometers [8]. Moreover, in reflective LSCI, photons detected are usually backscattered from parenchyma tissue after penetrating through the superficial microvessels. Because both the absorption and scattering of superficial microvessels are usually very weak, the dynamics of speckle pattern are mainly modulated by light that backscattered either from deeper parenchyma tissues or big vessels rather than directly from superficial microvessels. So, it is also hard to resolve the blood flow of microvascular in superficial tissue by reflective LSCI, as shown in Fig. 1(h).
Fig. 1. Dual-exposure temporal laser speckle blood flow image and angiography of mouse ear. (a) Flow chart of data acquisition and processing for dual exposure transmissive temporal LSCI. (b) Schematic diagram of transmissive laser speckle contrast imaging system. (c) Blood flow index ${1 / {{\tau _c}}}$ image. (d) Speckle angiographic $\beta {\xi ^2}$ image. (e) Fluorescence angiographic image. (f) ${1 / {K_t^2}}$ image of transmissive speckle temporal contrast image. (g) ${1 / {K_s^2}}$ image of transmissive speckle spatial contrast image. (h) The reflected speckle temporal contrast image. (i) Normalized intensity distribution of the selected LOIs indicated by the yellow lines in the Zoom-in images in (c)-(h). The black arrows and numbers in (i) correspond the blood vessels indicated by the white arrows and numbers in (c)-(h).
In biological tissues, Mie scattering is the predominant scattering mechanism. The probability of forward scattering is much higher than backward scattering when light propagates through tissues, and the number of scattering events is relatively low. Therefore, with transmission illumination, forward-scattered light not only penetrates much deeper into the tissue than backward-scattered light but also provides higher spatial resolution compared to reflective imaging [9]. Based on these fundamental advantages of transmissive imaging, Dunn et al. proposed to improve the detecting depth and spatial resolution of blood flow imaging by transmissive LSCI, which achieve centimeter penetration in human finger joint synovial and collateral artery in mouse hindlimb [10,11]. Li and Wang presented an approach to achieve microangiography with a transmissive laser speckle imaging system through statistical analysis of dynamic scattering signal separated from static scattering signal by utilizing an eigen-decomposition filter [12]. By substituting the original light source with a low coherence light emitted diode (LED), they realized a velocity mapping of erythrocyte flow in microvascular by correlation analysis of dynamic signal that removed background signal [13]. However, this method can only measure the speed of selected points rather than the complete speed map, and it is difficult to obtain flow rate for large blood vessels. Recently, Li et al. also showed an improvement on Signal-to-Background Ratio of blood flow imaging in thick tissues by transmissive LSCI, such as mouse hindlimbs, human fingers and wrists [14]. The improvement on spatial resolution of blood flow in thin tissues such as mouse ears and dorsal skin was also validated. However, it should be noted that the forward propagating light is a mixture of scattering and nonscattering in transmissive imaging according to the fundamental of tissue optics. There will be a large amount of directly transmitted ballistic photons in the light propagating forward through a thin tissue with the thickness ranging from submillimeter to millimeter [15]. Previous studies of transmissive LSCI adopted the same theoretical model used in reflective LSCI to estimate the blood flow speed index without considering the contribution of nonscattered light. We demonstrated here that the usage of traditional laser speckle contrast analysis for transmissive laser speckle imaging could lead to overestimation of background flow velocity, especially for the low scattering thin tissues, where the estimated speed of the blood vessel position is lower than the background tissue, as shown in Fig. 1(f).
To solve this problem, we derived a theoretical model of transmissive temporal LSCI accounting for the component of nonscattered light. Based on the new model, we further proposed a dual exposure temporal transmissive LSCI method for simultaneously accessing angiography and blood flow speed at microvessel level. In vivo animal experiments clearly demonstrated that our method possesses excellent vascular resolution capability, can distinguish microvascular structures with a diameter of approximately 20 micrometers. Moreover, it allowed for the assessment of the dynamic response of microvascular morphology and blood flow velocity induced by drug. Furthermore, our method monitored angiogenesis and vascular remodeling during mouse ear tumor growth in real time, demonstrating the potential of this method for detecting and studying microcirculation. In conclusion, we have achieved the monitoring of dynamic changes of vascular morphology and blood flow velocity at the microvessel level through the application of a simple label-free wide-field imaging approach.
2. Materials and methods
2.1 Theoretical model of transmissive LSCI
When the transmissive geometry was adopted, the electric field received by the image sensor was set as ${\boldsymbol{E}_{}}(x,y,t)$, which was the superposition of electric field ${\boldsymbol{E}_{ns}}(x,y,t)$ of the directly transmitted light and the electric field ${\boldsymbol{E}_s}(x,y,t)$ of scattered light.
Therefore, the intensity autocorrelation function of transmissive light could be written as
To simplify the expression in Eq. (2), we set x, y, and t to zero. Then Eq. (2) can be simplified as
Here a new parameter $\xi = {{\left\langle {{I_s}} \right\rangle } / {\left( {\left\langle {{I_s}} \right\rangle + \left\langle {{I_{ns}}} \right\rangle } \right)}}$ was defined to represent the ratio of scattered light intensity to the total transmitted light. ${g_{2,s}}(\Delta x,\Delta y,\tau )$ and ${g_{2,ns}}(\Delta x,\Delta y,\tau )$ were the intensity spatiotemporal autocorrelation function of scattered and nonscattered light respectively. ${g_{1,s}}(\Delta x,\Delta y,\tau )$ and ${g_{1,ns}}(\Delta x,\Delta y,\tau )$ represent the electric field correlation functions of scattered and nonscattered light respectively. The simplification from Eq. (2) to Eq. (3) is detailed in Eq. (S1) to Eq. (S3) in Supplement 1. The intensity autocorrelation function can be related to the electric field correlation function by the Siegert relation:
The electric field autocorrelation function can be decomposed into spatial autocorrelation ${g_{1,S}}(\Delta x,\Delta y)$ and temporal autocorrelation functions. Notably, the spatial autocorrelation function differs between scattered and nonscattered light. As the electric field of directly transmitted light is predominantly modulated by spatially varying tissue absorption along the propagation path and exhibits distinct absorption characteristics at different tissue locations, the electric field of non-scattered light becomes spatially uncorrelated. Thus, the spatial autocorrelation function of nonscattered light was zero, ${g_{1,S}}(\Delta x,\Delta y) = 0$. Consequently, ${g_{1,ns}}(\Delta x,\Delta y,\tau ) = 0$ and ${g_{2,ns}}(\Delta x,\Delta y,\tau ) = 1$ in Eq. (4). Substituting into Eq. (3) gives the intensity autocorrelation function of transmissive light:
where $\beta$ was the system correction factor, reflects the effects of the source coherence properties and the spatial and temporal averaging of the speckle dynamics and other factors [1]. The formula and physical interpretations of $\beta$ was provided in Eq. (S4). Therefore, the second-moment of light intensity of the photoelectric field received by the detector can be written as [16]:Here, T was the exposure time. The form of the field correlation function ${g_{1,s}}(\tau )$ is generally defined as [17,18]:
Here, ${\tau _c}$ was the decorrelation time of the electric field, inversely proportional to the velocity of the scattering particles. The parameter ($n$) varies depending on the dynamic light scattering regime and the type of motion for the light scattering particles. In our study, we primarily adopt the Lorentzian line shapes [$n = 1$ in Eq. (7)] [19,20], as red blood cells in microvessels predominantly undergo directional movement, and photon scattering from these cells is primarily characterized by multiple scattering.
Thus, considering the directly transmitted nonscattered component, the temporal contrast $K_t^2$ of the transmissive laser speckle could be written as:
In the traditional model of LSCI, the expression of $K_{}^2$ is written as [21]:
Compared with the traditional model of LSCI, the expression of laser speckle temporal contrast in new model was multiplied by a coefficient ${\xi ^2}$ reflecting the scattering level of tissue after the nonscattered ballistic light was considered.
2.2 Dual exposure temporal laser speckle contrast imaging
When the exposure time T was short enough, based on Eq. (8), the temporal laser speckle contrast can be estimated as:
So, the scattering level of tissue, $\beta {\xi ^2}$, can be approximated as $K_{t,short}^2$ obtained by calculating the squared temporal speckle contrast of laser speckle images recorded with a short exposure time. Considering the minimum exposure time of the commonly used commercial CCD/CMOS camera is usually in the range of tens of microseconds, here the exposure time of T = 21 µs was used for the imaging as short exposure time here. In practice, time contrast is defined as the ratio of the standard deviation to the mean of an image sequence. Spatial contrast is defined as the ratio of the standard deviation to the mean of the pixels in a 5 × 5 sliding window. The speckle size is approximately 1.2 pixels, is calculated based on the spatial displacement autocorrelation of the speckle pattern obtained from transmission detection geometry.
The imaging system as well as the flow chart of data acquisition and processing was shown in Figs. 1(a) and 1(b). Firstly, $\beta {\xi ^2}$ was approximated as the squared temporal contrast $K_{t,short}^2$ of a series of laser speckle images recorded with a short exposure time. Then a series of laser speckle images were acquired at a long exposure time (T = 20 ms), and the temporal contrast $K_{t,long}^2$ was calculated by substituting the $\beta {\xi ^2}$ and $K_{t,long}^2$ into Eq. (8). The decorrelation time ${\tau _c}$ of transmitted light field could be resolved by Newton iteration. Since the speed of scattering particles was known to be inversely proportional to ${\tau _c}$, ${1 / {{\tau _c}}}$ was used as the blood flow index to quantify the blood flow speed in flowmetry.
2.3 Image system
The dual exposure transmissive laser speckle contrast imaging system is shown in Fig. 1(b). A semiconductor laser (785 nm, 90 mW, Thorlabs, USA) was used as the linearly polarized illumination source. After collimated by a collimating lens (f = 30 mm, Thorlabs, USA), the laser beam irradiated on the sample through a reflector. The imaging system was a stereomicroscope (MVX10, Olympus, Japan), combined with an objective (MVPLAPO1X, NA 0.25, Olympus, Japan) and a tube lens (MVX-TV0.63XC, 0.63×, Olympus, Japan). The original laser speckle images were captured by a CMOS camera (acA2040-120um, Basler, Germany), and the acquisition frame rate was 50 fps.
The above image system was controlled by a custom-written software based on Labview, and the image acquisition process was shown in Fig. 1(a). Firstly, the neutral density filters (T = 0.032%, GCC-301071 and GCC-301031, Daheng Optics, China) were removed and 50 frames of speckle images were acquired with the exposure time of T = 21 µs. Then, the servo transferred the neutral density filters into the illumination path and 50 frames of speckle images were acquired with the exposure time of T = 20 ms. The above steps were repeated to realize time-series in vivo imaging. The neutral density filters ensure the same intensity of the images taken at the long and short exposure times. The $\beta {\xi ^2}$ images and ${1 / {{\tau _c}}}$ images were resolved by calculating the temporal contrast of the laser speckle images with short and long exposure.
For comparison, transmissive s-LSCI images and transmissive t-LSCI images were obtained from the spatial contrast and the temporal contrast calculated from 50 frames of laser speckle images with long exposure, respectively. The blood flow was approximated by ${1 / {{K^2}}}$.
Also, for comparison, traditional reflected LSCI were performed with the same imaging system by changing the position of the laser, and the blood flow was approximated by ${1 / {{K^2}}}$ calculated from 50 frames of laser speckle images.
2.4 Phantom preparation
To demonstrate that scattering components of tissue could be estimated by the transmissive temporal contrast of laser speckle image with short exposure, we conducted phantom validation, as depicted in Fig. 3(a). Intralipid 20% (Kelun, China) and pure water were mixed in various volume ratios to create an intralipid gradient dilution, ranging from 0.5% to 6% v/v with a gradient of 0.5%. The concentration of the intralipid solution was directly proportional to its scattering degree. In the experiment, a micro syringe controlled by a syringe pump (TJ-4A, LongerPump, China) injected different concentrations of intralipid solution into a capillary glass tube (inner diameter 100 µm) at a speed of 5 mm/sec.
To demonstrate that the speed of moving particles could be measured by the transmissive temporal LSCI, we conducted phantom validation, as depicted in Fig. 3(b). The agarose (Tsingke, China) and intralipid were mixed with pure water to create a mixture consisting of 2% v/v intralipid and 2% w/v agarose. The mixture was heated, dissolved, and stirred thoroughly before being poured into the mold. A capillary glass tube (inner diameter 100 µm) was embedded at a depth of 1 mm from the solution's surface and 2.5 mm from the bottom surface before the solution solidified. The gel was formed through cooling and solidification, and the reduced scattering coefficient of 2% v/v intralipid closely resembles that of biological tissue [22,23]. The capillary glass tube embedded in the gel simulated blood vessels in biological tissue. After anesthetizing the mice, blood was drawn from the heart, promptly injected into anticoagulant tubes with EDTA-2 K, and thoroughly mixed. In the experiment, the micro syringe filled with mouse blood was fixed on the syringe pump. The capillary glass tube on the gel and micro syringe were connected with PE tubing. Controlled by the syringe pump, the mouse blood was injected into the capillary glass tube at different speeds from 0 mm/sec to 20 mm/sec in 2.5 mm/sec increments.
2.5 Animal preparation
All experiments were approved by the Ethics Committee of Huazhong University of Science and Technology and efforts were made to minimize the number of animals used. Mice were anaesthetized by intraperitoneal injection of a mixed solution of chloral hydrate (200 mg/kg) and ethyl carbamate (1 g/kg) in saline according to body weight. Depilation cream was used to depilate the imaging area. For the dorsal skin imaging experiment, the dorsal skin was cut with surgical scissors and fixed to the slide for imaging. In the fluorescence angiography, TRITC-Dextran 70 (70 kDa, 1%, 7.5 ml/kg, Tdb lab, Sweden) in normal saline was intravenous injected.
To dynamically monitor the microvascular responses induced by ACh, the C57BL/6 mice were anesthetized and the imaging area was depilated. During the experiment, a group of dual exposure images was taken every 5 seconds, including 2 minutes of baseline images and 8 minutes after the intravenous injection of 120 µl ACh solution (0.25 mg/ml, Sigma-Aldrich, USA).
In order to establish a transplanted tumor model on mouse ear, the breast carcinoma cells EO771-mCherry were suspended in PBS (2.5 × 104 /µl) and then subcutaneously inoculated 20 µl mixture into the mice (C57BL/6, female, 7 weeks old) ear. The inoculation date was set as Day 0. Transmissive LSCI and tumor fluorescence imaging were performed on Day 2, 4, and 8. The fluorescence of EO771-mCherry was excited at 540 nm and emitted at 605 nm. The speckle angiographic images were binarized by Ostu threshold to analyze the vessel density.
2.6 Data analysis
We computed blood flow map and angiography from speckle imaging using MATLAB (R2017a). In order to compare the spatial resolution of microvessels by different imaging methods, a line of interest (LOI) was selected at the same position of each image. The normalized intensity distribution along the LOI was obtained. The vascular resolution of the imaging modality was determined based on the full-width at half-maximum (FWHM) of the smallest vessels detectable according to the intensity curve. In the monitoring results of vascular structure and functional responses induced by ACh in mouse ear, blood vessel structure images were obtained by applying Otsu thresholding to the blood flow images collected at different time points. Nine equidistant lines were drawn vertically along the centerline of the vessels in the vascular structure images. The boundaries of the vessels were automatically detected based on the sharp changes in pixel intensity along each line, and the vessel diameter is the mean of the diameters measured along the nine LOIs. In the mouse ear tumor experiment, tumor fluorescence images and laser speckle blood flow images were registered and fused. Subsequently, two regions of interest (ROIs) were selected within the microvascular area around the tumor. The vascular density and blood flow velocity changes within these ROIs were then calculated.
To assess the accuracy of blood flow velocity estimation, we conducted a comparative analysis between traditional transmissive LSCI and our novel dual-exposure transmissive LSCI, employing the Flow-to-Background Ratio (FBR). The FBR is defined as the ratio of the disparity between the estimated velocity of the blood vessel and the background, to the estimated velocity of the blood vessel itself. A negative FBR value signifies an overestimation of the flow velocity in the background, a value of 0 indicates complete indistinguishability between the background and blood vessels, and a value of 1 signifies the complete distinguishability between the background and blood vessels. ROIs were selected both on the background and vessels, outlined as white rectangular areas in Fig. 3(e) and (f). Statistical analysis of the FBR distribution at various flow velocities was carried out for both imaging methods (n = 5). Paired Student’s t-tests were executed on each dataset using Prism GraphPad software, and exact two-tailed P values were calculated with a significance level set at 0.05. The data are presented as mean ± standard deviation of the mean (SD).
3. Results
3.1 Theoretical analysis and phantom validation
According to the theoretical model of transmissive temporal LSCI shown in Eq. (8), the ${1 / {K_t^2}}$ of transmissive t-LSCI not only depended on the decorrelation time ${1 / {{\tau _c}}}$ related to the moving speed of the scattering particles, but also on the proportion of the nonscattered components in the transmitted light (Fig. 2). For tissues without nonscattered components that correspond the case of $\xi $=1, the expression of temporal laser speckle contrast follows the traditional theoretical model of LSCI. As the increase of the contribution of nonscattered components in transmitted light, $\xi $ decreases and ${1 / {K_t^2}}$ increases for a given ${T / {{\tau _c}}}$, which suggests that the speed of moving scattering particles estimated by ${1 / {K_t^2}}$ will vary with the scattering level of tissue. In another word, ${1 / {K_t^2}}$ will be significantly overestimated for the low-scattering tissue if the contribution of nonscattered component in the transmitted light was not accounted for in the model of temporal LSCI.
According to Eq. (8), $\beta {\xi ^2}$ can be estimated by the transmissive temporal laser speckle contrast obtained using a short exposure time. To validate the estimation ability of our method for scattering degree, we devised the phantom experiments as shown in Fig. 3(a). A series of intralipid solutions with varying scattering degrees were injected into the capillary glass tube at the same speed, and transmissive temporal LSCI was performed within a short exposure time of 21 µs. The concentration of the intralipid solution exhibited a direct proportionality to its scattering degree. As shown in Fig. 3(c), the measured $\beta {\xi ^2}$ increased with the rise in intralipid concentrations, aligning with the theoretical derivation.
Fig. 3. Blood-intralipid phantom validation. (a) Schematic diagram illustrating the phantom for estimating the scattering level using $\beta {\xi ^2}$. (b) Schematic diagram illustrating the phantom for estimating the flow speed of blood fluid. (c) $\beta {\xi ^2}$ images of the phantom used in (a) with different intralipid concentrations and the plot of $\beta {\xi ^2}$ within the white rectangular ROIs. (d) $\beta {\xi ^2}$ images of the phantom used in (b), along with the plots of $\beta {\xi ^2}$ within the white rectangular ROIs. (e) Blood flow index ${1 / {{\tau _c}}}$ images of the phantom used in (b), the plots of ${1 / {{\tau _c}}}$ within the white rectangular ROIs and the profiles of ${1 / {{\tau _c}}}$ indicated by the white line. (f) Transmissive speckle temporal contrast image of the phantom used in (b), along with the plot of ${1 / {K_t^2}}$ within the white rectangular ROIs and the profiles of ${1 / {K_t^2}}$ indicated by the white line. (g) Reflected speckle temporal contrast image of the phantom used in (b), along with the plot of ${1 / {K_t^2}}$ within the white rectangular ROIs and the profiles of ${1 / {K_t^2}}$ indicated by the white line. (h) Flow-to-Background Ratio comparison between the dual-exposure t-LSCI method and traditional transmissive t-LSCI at various flow rates.
In order to correct the estimation error of motion speed caused by the nonscattered components of tissues, we proposed the dual exposure transmissive temporal LSCI. This method firstly estimated the scattering components $\beta {\xi ^2}$ of the tissues by the transmissive temporal LSCI with short exposure time. Then ${1 / {{\tau _c}}}$ was reconstructed according to Eq. (8) and $\beta {\xi ^2}$ by using the speckle images recorded with long exposure time. To validate the capability of our dual-exposure transmissive temporal LSCI for angiography and blood flowmetry, we devised the phantom experiments as shown in Fig. 3(b). In this setup, we injected mouse whole blood into a capillary glass tube embedded in an intralipid-agarose gel background at varying speeds, followed by the execution of double-exposure transmission laser speckle imaging. As shown in Fig. 3(d), the scattering component estimated by $\beta {\xi ^2}$ of the background intralipid-agarose gel was smaller than that of whole mouse blood used in our experiments. As shown in Fig. 3(e), the ${1 / {{\tau _c}}}$ in the region of moving mouse blood increased linearly with the increase of flow speed in the range from 0 to 20 mm/s, and the value in stationary background intralipid-agarose gel consistently approach zero. It can be clearly seen in Fig. 3(f) that using ${1 / {K_t^2}}$ to estimate blood flow under the transmission detection geometry result in overestimation of the background intralipid-agarose gel. In Fig. 3(h), the stability of the FBR near 1, as observed with varying flow velocities, validates the robustness of our new method for blood flow estimation. This indicates the method's capability to accurately estimate flow velocities within the range of 0-20 mm/s and effectively discern between the flow region and background. Additionally, Fig. 3(h) validates the vulnerability of traditional transmissive laser speckle temporal contrast imaging (t-LSCI) for blood flow estimation. In the 0-10 mm/s range, the FBR is less than 0, suggesting an overestimation of background by traditional LSCI analysis method. In the velocity range of 10-20 mm/s, the FBR approaches 0, suggesting that traditional LSCI analysis methods estimate similar velocities for both the flowing region and the background within this range. This similarity poses a challenge in distinguishing between the static background and the flowing region [Fig. 3(h)]. Using ${1 / {K_t^2}}$ to estimate blood flow under the reflection detection geometry also exhibits a limitation, where the ${1 / {K_t^2}}$ in the blood region appears lower than that in the background, especially noticeable when the blood flow velocity is low [Fig. 3(g)]. This phenomenon is also evident in the in vivo imaging results of the mouse ear [see Fig. 1(h), indicated by the red arrow].
3.2 Microvascular speckle angiography and blood flowmetry
By performing the dual exposure transmissive temporal laser speckle contrast imaging in mouse ear, blood flow speed and blood vessel morphology at capillary level could be simultaneously mapped. $\beta {\xi ^2}$ image [Fig. 1(d)] achieved the label-free speckle angiography of microvessels by differentiating the scattering level of blood vessel and nonvascular tissues using the ratio of scattered light intensity to the total transmitted light intensity, which can be demonstrated in the fluorescence angiography [Fig. 1(e)] obtained by fluorescence imaging. The blood flow index ${1 / {{\tau _c}}}$ image [Fig. 1(c)] demonstrated higher spatial resolution of blood flow speed obtained by our method than that by transmissive laser speckle spatial contrast imaging (s-LSCI) [Fig. 1(g)] and reflected t-LSCI [Fig. 1(h)]. The normalized intensity profiles of the LOIs in above images were shown in Fig. 1(i). We can see that the blood flow index image and speckle angiographic image resolved by our new transmissive temporal LSCI distinguish the microvessels with a diameter of less than 20 µm. The similar results were also obtained from other tissues with the thickness of hundred micrometers, such as mouse dorsal skin (see Fig. S2 in Supplement 1).
In LSCI, $1/K_t^2$ is usually used as the approximation of blood flow perfusion [24,25]. When the contributions of nonscattered components were not considered, the values of ${1 / {K_t^2}}$ in microvessels were significantly lower than those of nonvascular background tissue, while the ${1 / {K_t^2}}$ of big blood vessel was higher than that of nonvascular background tissue, as shown in Fig. 1(f). It was worth noting that in traditional reflected t-LSCI, there was also the phenomenon that ${1 / {K_t^2}}$ of microvessels was lower than that of nonvascular tissue, as indicated by the red arrow in Fig. 1(h). It can further be seen in Fig. 1(i), the normalized intensity along the LOI showed the peaks of ${1 / {{\tau _c}}}$ plot at the position of blood vessel for our new dual- exposure LSCI method, whereas the valleys or irregular changes appeared in the plot of ${1 / {K_t^2}}$ at the position of blood vessel for traditional reflected t-LSCI and transmissive t-LSCI. The overestimation of the ${1 / {K_t^2}}$ in nonvascular background tissue in traditional temporal LSCI model not only introduced significant errors in the quantitative analysis of blood flow, but also made it difficult to perform the insight image analysis of blood flow such as the segmentation of blood vessel, parameters analysis of vascular morphology, classification of diseases using blood flow images.
Our new model corrected the overestimation of blood flow index in nonvascular background tissue [Fig. 1(c)], through separating the contributions of nonscattered components from the transmitted light field. Besides, the spatial resolution of ${1 / {{\tau _c}}}$ image was higher than that of blood flow images resolved by reflected t-LSCI [Fig. 1(h)] and transmissive s-LSCI [Fig. 1(g)].
3.3 Spatio-temporal changes in blood flow and angiography in microvessels
Microvascular responses induced by ACh were observed by our dual-exposure transmissive temporal LSCI. The spatio-temporal dynamic changes of vascular morphology and blood flow were shown in Fig. 4. Investigating the changes of speckle angiography images [Fig. 4(a)], the diameter of the blood vessel indicated by the white arrow in Fig. 4(a) dilated to 196% of the baseline after intravenous administration of ACh [Fig. 4(g)]. The ${1 / {{\tau _c}}}$ image [Fig. 4(b)] and the time courses of changes in blood flow [Figs. 4(e) and 4(f)] showed that the intravenous administration of ACh firstly induced a rapid decrease of blood flow in microvessels of mouse ear, followed by the hyperperfusion and recovery to the baseline.
Fig. 4. Vascular functional response induced by ACh in mice ear. (a) Speckle angiographic images $\beta {\xi ^2}$, (b) blood flow index images ${1 / {{\tau _c}}}$, (c) ${1 / {K_t^2}}$ images of transmissive t-LSCI, and (d) ${1 / {K_s^2}}$ images of transmissive s-LSCI at different time points before and after intravenous injection of ACh, the zoom-in views of the ROI were located at the bottom of each image. (e) Blood flow responses in the big vessel indicated by the yellow arrow in (b)-(d). (f) Blood flow responses in the microvessel indicated by the purple arrow in (b)-(d). (g) The changes of diameter of blood vessel indicated by the white line in (a). (h) The intensity profiles of blood vessel indicated by the white line in (a).
Spatio-temporal changes in blood flow induced by ACh validated the robustness of our dual exposure transmissive t-LSCI method on observing the changes in microvascular morphology and flowmetry. It is worthy to note that the values of in nonvascular background tissues are significantly overestimated when the transmitting light contains nonscattered component. In this case, it may lead to the failure of investigating the vascular morphology if the images are used to visualize the blood vessels as usually done in the traditional t-LSCI method. As shown in the vessel indicated by the black arrow in Fig. 4(c) and the intensity profiles of these vessels in Fig. 4(h), the distinction between blood vessels and nonvascular tissue becomes challenging during blood flow changes, and the results reveal the vulnerability of the traditional transmission t-LSCI. During the rapid decrease in blood flow from 0 s to 35 s after intravenous administration of ACh, the ${1 / {K_t^2}}$ in blood vessel was lower than that in nonvascular background tissues allowing for clear differentiation. However, as blood flow increased subsequently, the ${1 / {K_t^2}}$ values in blood vessels gradually became comparable to that in nonvascular background tissue, which made it difficult to extract vessel information from the background [Figs. 4(c) and 4(h)]. This phenomenon can also be found in Fig. 3(h). Our new dual exposure transmissive temporal laser speckle contrast imaging method successfully obtained the dynamics of blood flow speed at microvessels. For the big vessels, the changes in blood flow could be obtained by both the traditional and our new temporal LSCI model. But, the time-varying change of blood flow in microvessels could be only shown by the blood flow index ${1 / {{\tau _c}}}$ images [Figs. 4(e) and 4(f)]. Although the transmissive s-LSCI [Fig. 4(d)] can reflect the changes in blood flow speed, it was limited to the big vessels whereas failed to resolve the blood flow changes in microvessels.
As shown in Fig. S3 in Supplement 1, after the intravenous administration of ACh, microvascular blood flow of mouse dorsal skin also showed a triphasic change of falling, then rising and finally returning to baseline. Similarly, compared with the transmissive t-LSCI [Fig. S3(c)] and the transmissive s-LSCI [Fig. S3(d)], the ${1 / {{\tau _c}}}$ image [Fig. S3(b)] resolved by our new transmissive temporal LSCI model not only ensured the accuracy of estimation of blood flow, but also the spatial resolution.
3.4 Angiogenesis and blood flow changes during tumor growth
The dual exposure temporal transmissive LSCI was applied to observe the angiogenesis and blood flow redistribution during the growth of transplanted tumor. The mouse ear transplanted tumor model was established by subcutaneously inoculating EO771-mCherry breast carcinoma cells into the mouse ear. Before and after the inoculation of tumor cells, mouse ears were imaged on the Day 0, Day 2, Day 4, and Day 8. The speckle angiographic images [Fig. 5(a)] and blood flow index images [Fig. 5(b)] were resolved by using our new transmissive LSCI model. The image obtained on Day 0 showed the vascular structure and blood flow index of the mouse ear without the transplanted tumor. The fluorescence of tumor was merged with the speckle angiographic $\beta {\xi ^2}$ images and the blood flow index ${1 / {{\tau _c}}}$ images. The tumor location was indicated by the gray arrows in the green area of Fig. 5(a) and the red area of Fig. 5(b).
Fig. 5. Angiogenesis and blood flow monitoring during tumor growth. (a) Speckle angiographic images merged with tumor fluorescence (green pseudocolor) of mouse ears. (b) Blood flow index images merged with tumor fluorescence (red pseudocolor) of mouse ears. In the zoom-in view of (a) and (b), the white arrows indicate neovascularization; the gray arrows indicate the tumor location. (c) Changes in vessel density and (d) changes in blood flow in the ROIs selected in zoom-in view of (a) and (b). Day 0: before tumor inoculation. Day 2, Day 4 and Day 8: after tumor inoculation.
Two days after tumor inoculation, the development of vascular changes around the tumor was shown in the images of speckle angiography and blood flow index. The zoom-in view of area indicated by the gray rectangular around the tumor was used to observe the changes of blood flow and angiogenesis during tumor growth. Two ROIs were selected for quantitative analysis of the changes of blood vessel density and blood flow [Figs. 5(c) and 5(d)]. With the growth of tumor, the blood flow speed of the vessels surrounding the tumor in the mouse ear increased [Figs. 5(b) and 5(d)], and neovascularization with fast blood flow speed appeared [indicated by the white arrows in Figs. 5(a) and 5(b)]. There was a high degree of vascular heterogeneity at the edge of the ear far from the tumor [outside the gray rectangular in Fig. 5(b)], with high blood flow density close to the periphery of the tumor, which may contribute to tumor growth, and farther regions with low blood flow density. During the period of neovascularization, an abundant vascular network around the tumor characterized by dilated, tortuous and disorganized was formed, which was clearly distinguishable compared to the original vasculature in Day 0. From Day 4 to Day 8, the tumor core became progressively hypovascular, and vessels in the vicinity of the tumor disappeared on Day 8, which may be due to the necrosis of the tumor core [26,27].
It should be noticed that there was an overestimation of blood flow in the tumor and the nonvascular background tissues if we used the ${1 / {K_t^2}}$ as the approximation of blood flow through the traditional transmissive t-LSCI model [see Fig. S4(a) in Supplement 1]. In Fig. S4(a), the ${1 / {K_t^2}}$ values of tumor and nonvascular background tissues were much higher than the ${1 / {K_t^2}}$ of the blood vessel around tumor. The blood flow index image of ${1 / {{\tau _c}}}$ obtained by our new model corrected these errors by considering the contribution of nonscattered components. Transmissive s-LSCI image also showed neovascularization during tumor growth [indicated by a white arrow in Fig. S4(b) in Supplement 1], as well as a trend of increasing and then decreasing blood flow speed and vessel density, which was consistent with that of ${1 / {{\tau _c}}}$ image obtained by our new method [Fig. 5(b)]. However, transmissive s-LSCI could not resolve the blood flow changes in microvessels surrounding the tumor like the blood flow index ${1 / {{\tau _c}}}$ did because of the sacrifice of spatial resolution.
4. Discussion
In this paper, we proposed a theoretical model and a dual exposure imaging method of transmissive temporal LSCI to simultaneously realize the label-free wide-field imaging of angiography and blood flowmetry at microvessel level.
It is worthy to note that in traditional transmissive t-LSCI image [see Fig. S4(a) in Supplement 1], the ${1 / {K_t^2}}$ values in the tumor core were higher than that in the surrounding tissues. This observation contradicts the hypoperfusion typically observed in the tumor core in previous studies [28,29]. However, as shown in Fig. 2, when the speed of scattering particles was kept constant, ${1 / {K_t^2}}$ of low scattering tissue was higher than ${1 / {K_t^2}}$ of the high scattering tissue. This suggests that the overestimated ${1 / {K_t^2}}$ of the tumor may be attributed to the lower scattering level of the tumor core compared to the surrounding tissue. Theoretical analysis and phantom experiments have proved $\beta {\xi ^2}$ is related to the tissue scattering level. It is also shown in Fig. 5(a) that the $\beta {\xi ^2}$ values in the tumor are lower than those in surrounding tissue or blood vessels. Our new method, by correcting the influence of the nonscattered component of transmitted light, provides a more accurate estimation of blood flow and confirms the presence of necrosis in the tumor core in the reconstructed $1/{\tau _c}$ images.
In reflected LSCI, the statically scattered light from stationary tissues seriously degrades the accuracy of flow speed estimation [30], a series of methods were proposed to correct the effects of static scattering components on velocity measurement [31–33]. For transmissive LSCI, the transmitted photons received by the detector penetrate the entire tissue, and there is a near absence of photons undergoing only static scattering. Therefore, there are few static components in the transmitted light field, unlike those present in the reflected light field. In our imaging model here, we did not consider static scattering components. We attempted to use a multi-exposure imaging model or incorporate static scattering components into the imaging model. However, this work currently faces challenges in model establishing and analytical solution solving. In the future, we will continue to explore the effects of static scattering in transmitted laser speckle flow imaging to further improve the model.
The correct form of the scattered electric field autocorrelation function is crucial for accurately determining the electric field decorrelation time ${\tau _c}$ [34]. Considering that red blood cells in microvessels predominantly exhibit directional movement, and the scattering resulting from photon-red blood cell collisions is primarily multiple scattering. Therefore, we chose the Lorentzian line shapes for ${g_{1,s}}(\tau )$ in our model. Given that some studies recommend using ${g_{1,s}}(\tau ) = \exp (\sqrt { - \tau /{\tau _c}} )$ to investigate microvessels with diameters <30 µm [18], we also employed this form of autocorrelation function in our model, resulting in the blood flow index image shown in Fig. S1(b). When ${g_{1,s}}(\tau ) = \exp (\sqrt { - \tau /{\tau _c}} )$, the decorrelation time of small vessels is primarily influenced [Figs. S1(c) and (d), black arrows 1 and 2], which shows a reduction in the blood flow index of the small vessels compared to ${g_{1,s}}(\tau ) = \exp ( - \tau /{\tau _c})$. The use of this form of ${g_{1,s}}(\tau )$ does not achieve the desired effect and proves less favorable for microvessel identification. This discrepancy may be attributed to the difference in the number of scattering events and the type of particle motion detected by transmission and reflection geometries.
In the traditional reflected LSCI theory, ${1 / {{K^2}}}$ was used as an approximation of $1/{\tau _c}$ [30]. According to the theory of dynamic light scattering, ${\tau _c}$ was inversely related to the scattering angle [16,35], as shown in Fig. S5. Our results show that the value of ${1 / {K_t^2}}$ resolved by the transmission detection geometry is smaller than that by the reflection detection geometry [see Fig. S2(a)]. Moreover, in the case of directional flow with the same velocity, the value of $1/{\tau _c}$ resolved by the transmission detection geometry [see Fig. 3(e)] is smaller than that by the reflection detection geometry in Ref. [30]. This difference is consistent with the multiple change of blood flow index corresponding to the difference of scattering angle of different detection geometry, as shown in Fig. S5. Furthermore, as depicted in the in vivo imaging of mouse ear [Fig. 1(h)], reflected t-LSCI could not accurately estimate the blood flow distribution of microvessels in which the values of ${1 / {K_t^2}}$ are also smaller than that of nonvascular background tissue. This discrepancy may be attributed to the inclusion of incompletely scattered photons in the backscattered light from low scattering thin tissues. Therefore, the traditional reflected t-LSCI, which is widely used in clinical applications, also warrants further optimization.
The label-free speckle angiography resolved the vascular morphology by differentiating the scattering level of transmitted light in vascular from that in nonvascular tissues. In this paper, we selected various mouse tissues with different thickness, including ears, dorsal skin, hind paw, hind limb and front paw, as imaging objects to evaluate the validity of our method. For micrometer-thick tissues, the results demonstrate the excellent ability of our method to simultaneously resolve microvascular morphology and blood flow speed in different application scenarios. For millimeter-thick tissues, we also showcase the potential of our method in monitoring blood flow in thicker tissues. Although the high thickness of millimeter-thick tissues results in almost total scattering of transmitted photons from both vessels and tissues, which limits the ability of $\beta {\xi ^2}$ image to resolve microvascular morphology. The ${1 / {{\tau _c}}}$ image provided equivalent resolution of blood flow as the transmissive t-LSCI images do, and were superior to the reflected t-LSCI images obtained by reflective geometry (see Fig. S2 in Supplement 1). In the future studies, NIR-II light with stronger penetration ability could be considered to replace the current light source, which could address the limitation imposed by high scattering in thick tissues. Additionally, we aim to further broaden expand the application of dual-exposure temporal laser speckle imaging method, and it will be further applied to clinical human parts with suitable thickness in the future.
Funding
National Natural Science Foundation of China (61890950, 61890951, 62275095, 82261138559); National Key Research and Development Program of China (2021YFC2400102); Innovation Project of Optics Valley Laboratory (OVL2021BG012); CAMS Innovation Fund for Medical Sciences (CIFMS, 2019-I2M-5-014); director fund of the Wuhan National Laboratory for Optoelectronics.
Disclosures
The authors declare no conflicts of interest.
Data availability
Data underlying the results presented in this paper are not publicly available at this time but may be obtained from the authors upon reasonable request.
Supplemental document
See Supplement 1 for supporting content.
References
1. D. A. Boas and A. K. Dunn, “Laser speckle contrast imaging in biomedical optics,” J. Biomed. Opt. 15(1), 011109 (2010). [CrossRef]
2. J. Senarathna, A. Rege, N. Li, et al., “Laser Speckle Contrast Imaging: theory, instrumentation and applications,” IEEE Rev. Biomed. Eng. 6, 99–110 (2013). [CrossRef]
3. S. M. S. Kazmi, L. M. Richards, C. J. Schrandt, et al., “Expanding applications, accuracy, and interpretation of laser speckle contrast imaging of cerebral blood flow,” J. Cereb. Blood Flow Metab. 35(7), 1076–1084 (2015). [CrossRef]
4. P. G. Vaz, A. Humeau-Heurtier, E. Figueiras, et al., “Laser Speckle Imaging to Monitor Microvascular Blood Flow: A Review,” IEEE Rev. Biomed. Eng. 9, 2392001 (2016). [CrossRef]
5. P. A. Dyachenko, A. N. Bashkatov, D. A. Alexandrov, et al., “Laser speckle contrast imaging for monitoring of acute pancreatitis at ischemia-reperfusion injury of the pancreas in rats,” J. Innov. Opt. Heal. Sci. 15(01), 2242002 (2022). [CrossRef]
6. J. W. Goodman, Speckle phenomena in optics: theory and applications (SPIE Press, 2007).
7. D. Briers, D. D. Duncan, E. Hirst, et al., “Laser speckle contrast imaging: theoretical and practical limitations,” J. Biomed. Opt. 18(6), 066018 (2013). [CrossRef]
8. M. A. Davis, S. M. S. Kazmi, and A. K. Dunn, “Imaging depth and multiple scattering in laser speckle contrast imaging,” J. Biomed. Opt. 19(8), 086001 (2014). [CrossRef]
9. N. G. Ferris, T. M. Cannon, M. Villiger, et al., “Forward multiple scattering dominates speckle decorrelation in whole-blood flowmetry using optical coherence tomography,” Biomed. Opt. Express 11(4), 1947–1966 (2020). [CrossRef]
10. J. F. Dunn, K. R. Forrester, L. Martin, et al., “A Transmissive Laser Speckle Imaging Technique for Measuring Deep Tissue Blood Flow: An Example Application in Finger Joints,” Lasers Surg. Med. 43(1), 21–28 (2011). [CrossRef]
11. J. K. Meisner, J. Niu, S. Sumer, et al., “Trans-illuminated laser speckle imaging of collateral artery blood flow in ischemic mouse hindlimb,” J. Biomed. Opt. 18(9), 096011 (2013). [CrossRef]
12. C. Li and R. Wang, “Dynamic laser speckle angiography achieved by eigen-decomposition filtering,” J. Biophotonics 10(6-7), 805–810 (2017). [CrossRef]
13. C. Li and R. Wang, “Velocity measurements of heterogeneous RBC flow in capillary vessels using dynamic laser speckle signal,” J. Biomed. Opt. 22(04), 46002 (2017). [CrossRef]
14. D. Y. Li, Q. Xia, T. T. Yu, et al., “Transmissive-detected laser speckle contrast imaging for blood flow monitoring in thick tissue: from Monte Carlo simulation to experimental demonstration,” Light: Sci. Appl. 10(1), 241 (2021). [CrossRef]
15. D. P. Wang and J. Xia, “Optics based biomedical imaging: Principles and applications,” J. Appl. Phys. 125(19), 191101 (2019). [CrossRef]
16. A. G. Mailer, P. S. Clegg, and P. N. Pusey, “Particle sizing by dynamic light scattering: non-linear cumulant analysis,” J. Phys.: Condens. Matter 27(14), 145102 (2015). [CrossRef]
17. D. D. Postnov, J. B. Tang, S. E. Erdener, et al., “Dynamic light scattering imaging,” Sci. Adv. 6(45), eabc4628 (2020). [CrossRef]
18. C. Liu, K. Kilic, S. E. Erdener, et al., “Choosing a model for laser speckle contrast imaging,” Biomed. Opt. Express 12(6), 3571–3583 (2021). [CrossRef]
19. K. Basak, G. Dey, M. Mahadevappa, et al., “Laser speckle imaging by adaptive contrast computation for microvasculature assessment,” Opt. Laser Eng. 62, 87–94 (2014). [CrossRef]
20. A. Khalil, A. Humeau-Heurtier, G. Mahé, et al., “Laser speckle contrast imaging: age-related changes in microvascular blood flow and correlation with pulse-wave velocity in healthy subjects,” J. Biomed. Opt. 20(5), 051010 (2014). [CrossRef]
21. R. Bandyopadhyay, A. S. Gittings, S. S. Suh, et al., “Speckle-visibility spectroscopy: A tool to study time-varying dynamics,” Rev. Sci. Instrum. 76(9), 093110 (2005). [CrossRef]
22. B. Aernouts, E. Zamora-Rojas, R. Van Beers, et al., “Supercontinuum laser based optical characterization of Intralipid(R) phantoms in the 500-2250 nm range,” Opt. Express 21(26), 32450–32467 (2013). [CrossRef]
23. S. L. Jacques, “Optical properties of biological tissues: a review,” Phys. Med. Biol. 58(11), R37–61 (2013). [CrossRef]
24. N. Li, X. F. Jia, K. Murari, et al., “High spatiotemporal resolution imaging of the neurovascular response to electrical stimulation of rat peripheral trigeminal nerve as revealed by in vivo temporal laser speckle contrast,” J. Neurosci. Methods 176(2), 230–236 (2009). [CrossRef]
25. J. C. Hong, Y. Wang, X. Chen, et al., “Fluctuations of temporal contrast in laser speckle imaging of blood flow,” Opt. Lett. 43(21), 5214–5217 (2018). [CrossRef]
26. R. Q. Lin, J. H. Chen, H. N. Wang, et al., “Longitudinal label-free optical-resolution photoacoustic microscopy of tumor angiogenesis in vivo,” Quant. Imaging Med. Su. 5, 23–29 (2015). [CrossRef]
27. H. C. Zhou, N. B. Chen, H. X. Zhao, et al., “Optical-resolution photoacoustic microscopy for monitoring vascular normalization during anti-angiogenic therapy,” Photoacoustics 15, 100143 (2019). [CrossRef]
28. C. Viallard and B. Larrivee, “Tumor angiogenesis and vascular normalization: alternative therapeutic targets,” Angiogenesis 20(4), 409–426 (2017). [CrossRef]
29. M. Sun, C. Li, N. Chen, et al., “Full three-dimensional segmentation and quantification of tumor vessels for photoacoustic images,” Photoacoustics 20, 100212 (2020). [CrossRef]
30. A. B. Parthasarathy, W. J. Tom, A. Gopal, et al., “Robust flow measurement with multi-exposure speckle imaging,” Opt. Express 16(3), 1975–1989 (2008). [CrossRef]
31. Y. Wang, D. Wen, X. Chen, et al., “Improving the estimation of flow speed for laser speckle imaging with single exposure time,” Opt. Lett. 42(1), 57–60 (2017). [CrossRef]
32. J. Hong, X. Zhu, J. Lu, et al., “Quantitative laser speckle auto-inverse covariance imaging for robust estimation of blood flow,” Opt. Lett. 46(10), 2505–2508 (2021). [CrossRef]
33. E. Hirst, O. Thompson, and M. Andrews, “Measurement of dynamic scattering beneath stationary layers using multiple-exposure laser speckle contrast analysis,” Dynamics and Fluctuations in Biomedical Photonics Ix 8580 (2013).
34. M. A. Davis, L. Gagnon, D. A. Boas, et al., “Sensitivity of laser speckle contrast imaging to flow perturbations in the cortex,” Biomed. Opt. Express 7(3), 759–775 (2016). [CrossRef]
35. E. T. Hanson, R. Borsali, and R. Pecora, “Dynamic light scattering and small-angle neutron scattering studies of ternary rod/coil/solvent systems,” Macromolecules 34(7), 2208–2219 (2001). [CrossRef]
