1.

Introduction

Many methods have been used to simultaneously measure blood flow and blood oxygenation changes. Dual-wavelength laser speckle contrast imaging (DW-LSCI) has shown to provide superficial blood oxygenation and flow changes, including using an optical chopper and a transistor-transistor logic modulated camera to monitor cerebral ischemia,¹ and a color camera to obtain blood oxygenation information.² Synchronized DW-LSCI was introduced with two coregistered, synchronized cameras to achieve faster acquisition rate, and has been applied to monitor blood flow and oxygenation changes in a stroke mouse model³ and a nerve-stimulation rat model.⁴

In order to acquire deep tissue blood flow and oxygenation measurements, combined diffuse correlation spectroscopy (DCS) and near-infrared spectroscopy (NIRS) systems have been used. Combined DCS–NIRS systems were used to monitor the cerebral hemodynamics of neonates with congenital heart defects,⁵ stroke patients,⁶ and critically brain-injured adults.⁷ Contact and noncontact dual-wavelength DCS systems have also been validated^8–10 and used in such applications as monitoring muscle hemodynamics during electrical stimulation,¹¹ head and neck tumors during radiation therapy,¹² cerebral ischemia,¹³ and muscle revascularization.¹⁴ However, DCS can be costly because it requires a correlator board to calculate the intensity autocorrelation function.^9,15

Diffuse speckle contrast analysis (DSCA) allows for deep tissue blood flow measurements and utilizes a point-source illumination, like DCS, but with the pixel analysis of speckle contrast, like laser speckle contrast analysis (LASCA).¹⁶ DCS, LASCA, and DSCA are all based on the temporal autocorrelation function,¹⁷ but DSCA has advantages in its expansion into imaging and tomography due to its nature of multiple source and detector without much modifications.^18,19

Spatial DSCA observes the speckle contrast ($K_{\mathrm{s}}$), defined as^20,21

Because DSCA uses near-infrared light to obtain deep tissue blood flow, CW NIRS measurements can be made with another laser source, to observe intensity fluctuations from two wavelengths. CW NIRS relates to optical density (OD), or the logarithmic ratio of baseline intensity and transient intensity values, to concentration changes of oxyhemoglobin (OHb) and deoxyhemoglobin (RHb) from two wavelengths^24,25

In this study, we propose an expansion of the DSCA system by modulating between two laser sources to acquire NIRS measurements. The mean intensity value of the speckle contrast calculation window was used as transient intensity value to calculate the NIRS data. The proposed DSCA–NIRS system requires just two different analysis methods of the speckle image to simultaneously observe deep tissue blood flow and blood oxygenation changes.

2.

System Development

Figure 1 shows the schematic of the combined DSCA–NIRS system. A $1\times 2$ optical switch (SW $1\times 2$, Sercalo, Switzerland) was used to alternate between two CW laser sources (temporal coherence $>10\mathrm{m}$) for 785 and 852 nm wavelengths (DL785-100-SO/DL852-100-SO, $\sim 100\mathrm{mW}$, CrystaLaser Technology). The optical switch was synchronized with the detecting CCD camera (CoolSNAP MYO CCD Camera, Photometrics) by trigger pulses generated by a data acquisition board (NI USB-6229 BNC, National Instruments). The operation of the CCD camera limited the optical switch operation to a temporal frequency of 2 Hz. The power emitted to the sample was approximately 8 mW for 785 nm and 10 mW for 852 nm. A $1\times 4$ fiber splitter ($1\times 4$ PLC fiber splitter, Fiberpia, Republic of Korea) was used to reduce the source’s power for safe use and one fiber propagated light into the sample. All the fibers were multimode fibers with a core diameter of $62.5\mu \mathrm{m}$. The detecting CCD camera was fixed with a biconvex lens (LB-1811-B, $f=35\mathrm{mm}$, Thorlabs Inc.) to focus the image. A magnification of $1:3$ allowed for an imaging width of 26 mm from the sample surface.

Fig. 1

Schematic of combined DSCA–NIRS system that alternates between the wavelengths of 785 and 852 nm to produce changes in OD and speckle contrast.

An in-house built LabVIEW Program (LabVIEW™ 2012, National Instruments) started the trigger signal to synchronize the optical switch and camera, acquire the image, and also controlled the exposure time, region of interest window, and gain. Speckle contrast and OD calculations were done in a MATLAB R2010b (MathWorks Inc., Natick, Massachusetts). The experiments were run with a 15-mm source–detector separation and processed with a $49\times 49$ window to increase the signal-to-noise ratio.²⁶

3.

Phantom Preparation

A flow phantom (${\mu }_{\mathrm{a}}=0.065{\mathrm{mm}}^{-1}$, ${\mu }_{\mathrm{s}}^{\prime }=1.015{\mathrm{mm}}^{-1}$ for 650 nm) was constructed similar to previous reports.^15,16,20 A silicone (Sylgard 184 silicone elastomer kit, Dow Corning) was mixed with India ink drops (Pro Art) and ${\mathrm{TiO}}_2$ powder (Junsei Chemical Co., Japan) for the desired optical properties. A transparent tube with an inner diameter of $1/{16}^{\prime \prime }$ was placed at a depth of 7.5 mm and filled with transparent beads to randomize the flow.^15,16,20 A single syringe pump (Harvard Apparatus) flowed 0.06% diluted^15,16,20 intravenous fat emulsion (Lipision 20%, JW Pharm., Republic of Korea) into the tube at flow rates from 0.02 to $0.14\mathrm{mL}/\mathrm{s}$. Assuming that capillary velocity is approximately $1\mathrm{mm}/\mathrm{s}$,¹⁷ the phantom flow rates cover the physiologically relevant velocities.

4.

Results

Figure 2 shows the average $1/K_{\mathrm{S}}^2$ value for the flow rates in increments of $0.02\mathrm{mL}/\mathrm{s}$ for 30 s. Exposure times of 0.5, 1, and 3 ms were tested to show the linear relationship between flow detection and camera exposure time.²² Figure 2 shows that the $1/K_{\mathrm{S}}^2$ value response becomes more linear with a shorter exposure time. The inlet graph also indicates a difference in $1/K_{\mathrm{S}}^2$ values for 3 ms exposure, within a closer range. These phantom flow results agree with LASCA findings,²⁷ which show that the sensitivity is lost to faster flow rates with longer exposure time, but for slower flow rates, which are more relevant to the approximate physiological range,¹⁷ there is larger difference in $1/K_{\mathrm{S}}^2$ values for all exposure times.

Fig. 2

Average $1/K_{\mathrm{S}}^2$ values (with standard error) for various flow rates at the exposure of 0.5 ms (blue triangle), 1 ms (red circle), and 3 ms (black square) for flow phantom test. The inlet graph shows a closer view of 3 ms exposure phantom results.

An arm occlusion protocol was also tested with the system, reviewed and approved by Gwangju Institute of Science and Technology’s Institutional Review Board (IRB 20140319-HR-10-01-02). The lens tube of the camera was placed on the forearm of the subjects and arterial occlusion was applied with an arm cuff inflated to 220 mm Hg of pressure on the bicep. Total hemoglobin (THb), as shown in Figs. 3Fig. 4–5, is the summation of OHb and RHb and is proportional to blood volume.²⁸ The calculated values were filtered with a moving average ($\text{window size}=5$).

Fig. 3

Relative OHb (solid red line), RHb (dotted blue line), THb (thick green line) concentration changes (left column) and $1/K_{\mathrm{S}}^2$ values (right column) for 785 nm (thick black line) and 852 nm (purple line) for arterial occlusion for in one subject at (a) 0.5 ms exposure and (b) 1 ms exposure.

Despite showing a more linear response at 0.5 ms exposure, we were unable to obtain consistent NIRS response at shorter exposure times (0.5 and 1 ms) for the arm occlusion protocol. Figure 3 shows an example of the issues—taken from one subject, for obtaining NIRS signal during arterial occlusion. This includes an underestimation of blood oxygenation changes at 0.5 ms and an unstable NIRS signal at 1 ms exposure time. However, the NIRS results were highly inconsistent between subjects and trials. There is an approximately 50% drop in $1/K_{\mathrm{S}}^2$ values for 1-ms exposure and 66% drop at 0.5-ms exposure. Both 0.5- and 1-ms exposure times show a recovery of approximately 150% of baseline $1/K_{\mathrm{S}}^2$ values. The drop and recovery are similar to spatial DSCA.¹⁶

In contrast, 3 ms exposure gave us consistent blood oxygenation changes from four subjects for NIRS and measured $1/K_{\mathrm{S}}^2$ changes during arterial occlusion. Figure 4 shows the average arterial occlusion oxygenation and flow changes from four subjects (male, ages 24 to 25 years old). Since the speckle contrast values were different between subjects, the $1/K_{\mathrm{S}}^2$ values were converted to percentage changes and then averaged to display. During arterial occlusion, we observed an approximate equivalent rise of RHb (maximum concentration change of $0.0085\pm 0.0024\mathrm{mM}/\mathrm{DPF}$) and fall of OHb (minimum concentration change of $-0.0057\pm 0.0044\mathrm{mM}/\mathrm{DPF}$), which follow trends similar to other studies.²⁹ For 3 ms exposure, the $1/K_{\mathrm{S}}^2$ values dropped to an average of $82.1\pm 4.0\%$ compared to baseline values, during the occlusion period, which is similar to that reported in temporal DSCA,²⁰ and the average recovery of $1/K_{\mathrm{S}}^2$ values after occlusion was $109.1\pm 0.8\%$, much less than the value from shorter exposure times.

For one subject, venous occlusion was ran to observe blood oxygenation and blood flow for different occlusion types (Fig. 5). Venous occlusion was achieved with 50 mm Hg of pressure. In contrast to arterial occlusion, venous occlusion showed a rise of just OHb during occlusion, with a slight drop in blood flow.³⁰ These results show the sensitivity of our system to differentiate between arterial and venous occlusion in terms of both blood oxygenation and blood flow measurements.

Fig. 4

(a) Relative OHb (solid red line), RHb (dotted blue line), THb (thick green line) concentration changes and (b) percentage change of $1/K_{\mathrm{S}}^2$ values for 785 nm (thick black line) and 852 nm (purple line) averaged from four subjects at 3 ms exposure, with standard error included for each graph. The occlusion period is marked by a gray bar in each of the graphs.

Fig. 5

Relative OHb (solid red line), RHb (dotted blue line), THb (thick green line) concentration changes (left axis) and $1/K_{\mathrm{S}}^2$ values for 785 nm (thick black line) and 852 nm (purple line) (right axis) for venous occlusion for one subject at 3 ms exposure time. The occlusion period is marked by a gray bar.

Figures 3–5 report $1/K_{\mathrm{S}}^2$ values from both wavelengths, 785 and 852 nm. There is not a large difference in measured $1/K_{\mathrm{S}}^2$ values between the two wavelengths. This finding gives us a confidence that alternating wavelengths for NIRS calculations does not affect DSCA calculations.

5.

Discussion and Conclusion

The proposed combined DSCA–NIRS system shows the capability to simultaneously measure qualitative blood flow and blood oxygenation by modulating between two laser sources and performing DSCA and NIRS analysis on light intensity fluctuations. However, the system still has limitations like the underestimation of blood flow drop (approximately 20% decrease from baseline) during the arterial occlusion compared to DCS results (50% or more of blood flow drop).^8,9,15,16 Therefore, the proposed combined DSCA–NIRS system may not suitable for quantitative measurements, yet is still applicable for investigating hemodynamic trends for physiological changes. The detector geometry (i.e., $\beta$ in DCS theory^16,20) and camera’s exposure time²² needs to be optimized in order to increase the sensitivity of the system.

Sensitive blood flow response with shorter exposure times and consistent blood oxygenation response at longer exposure times may be an inherent tradeoff for a combined DSCA–NIRS system. Whereas NIRS calculations depend on adequate mean intensity at the detector, DSCA requires sufficient speckle contrast to improve the sensitivity to flow rates. One solution to this tradeoff may be a multiexposure technique as it has already shown to offer quantitative information of the blood flow change³¹ and may be more sensitive in its NIRS measurements. Moreover, the current system must be validated by the concurrent measurements with validated techniques such as DCS to investigate the underestimation of blood flow change during the occlusion protocol and phantom experiments.^8,9,15,16,20