2.1.

LSF Setup and Data Analysis

When a coherent light source passes through a static tissue, it creates a random interference pattern known as speckle. The speckle pattern produced is characteristic for a given tissue structure. When particles move within the given tissue (i.e., red blood cells), it causes the speckle pattern to fluctuate, such that when these fluctuations are integrated over time it is possible to use spatial statistics to determine tissue motion.²¹ We have previously shown that, when using a simplified algorithm developed to remove the assumption of Gaussian or Lorentzian distributions,²² it is possible to find relative changes in blood flow in the microcirculation at an individual vessel level.¹⁸ The approach relates the change in speckle contrast [$C$, Eq. (1)] to autocorrelation time [$\tau$, Eq. (2)], which is inversely proportional to blood velocity [speckle intensity (SI), Eq. (2)]

In our arrangement, a 30-mW, 658-nm laser diode (LPM658-30, Newport Corporation, Irvine, CA) was coupled to a fiber optic cable and used to illuminate the tissue or tissue phantom from either an epi-illumination or trans-illumination orientation, where the fiber output was placed at the base of the (phantom) tissue. An illustration of the laser speckle imaging system, as applied to the mouse gracilis adductor collateral arteries in the trans-illumination orientation, is shown in Fig. 1. For both in vitro and in vivo testing, a cooled, monochrome charge-coupled device camera (Optronics Quantifier, Goleta, CA) ($7.4\times 7.4{\mu \text{m}}^2\text{pixels}$) was used to acquire the raw speckle images through an intravital microscope (Zeiss Axioskop, Thornwood, NY) using a $4\times$ air objective (Zeiss Acroplan LD $\mathrm{NA}=0.1$). As previously described,¹⁸ the objective and camera were chosen to balance the resolution of the scope with the size of the speckle to satisfy the Nyquist sampling criteria of at least 2 pixels per individual speckle.²³

Fig. 1

In vivo intravital imaging of collateral artery hemodynamics. The fiber optic cable was positioned for either trans-illumination or epi-illumination. For the trans-illumination configuration, the fiber optic outlet was placed on the opposite side of the leg from the exposed muscle and against the depilated skin to yield an evenly illuminated field of view. For the epi-illumination configuration, the fiber optic outlet was positioned to produce an unobstructed and even sheet of light on the surface of the exposed muscle. The given field of view was shifted using an $xy$-stage.

For each field of view, laser position, and flow setting, a sequence of $\ge 15$ 12-bit raw speckle images was acquired with 5-ms exposure time to capture average velocity over multiple cardiac cycles. Camera gain and light path intensity were varied to maintain mean pixel intensity of the raw speckle image. To minimize the influence of whole background tissue motion from mouse movement, the processed SI images were normalized to median background intensity. Code for converting raw speckle images into speckle flow maps was developed, implemented, and analyzed in Fiji,²⁴ and is freely available upon request. Individual flow images were then merged into larger two-dimensional maps using Adobe Photoshop (CS2, Adobe Systems Inc., San Jose, CA).

2.2.

Tissue Phantom Studies

While initial studies suggested that speckle signal analysis from transmitted laser light yields dominant vascular signals from deeper tissue structures²⁰ and that speckle signals remain viable through up to 2 cm of biological tissue,²⁵ the applicability of signal processing used for epi-illumination versus trans-illumination was unknown. Therefore, light scattering tissue phantoms were used to test speckle velocity linearity between trans-illumination and epi-illumination speckle arrangements. Phantoms were generated as previously described¹⁸ by adding titanium dioxide nanopowder (677469, Sigma-Aldrich, St. Louis, MO) to 184 Sylgard PDMS gel (Dow Corning, Midland, MI) at $10\mathrm{mg}/\mathrm{mL}$ and pouring the mixture over optical glass-bottomed molds. Phantoms were then allowed to cure overnight at 37°C. Molds 5-mm thick were composed of polyethylene tubing connected to glass capillary tubes with 142-μm inner diameter (Drummond Microcap, P1799 Sigma-Aldrich, St. Louis, MO) placed at 0, 120, 360, and 720 μm from the surface. These values were chosen to approximately represent the range of depths at which the collateral arteries reside under the top surface of the mouse gracilis muscle, which has a thickness of $\sim 750\mu \text{m}$.

For validating the linear operation of the speckle processing algorithm for both epi-illumination and trans-illumination arrangements, a sequence of known flows of defibrinated bovine blood (Hemostat Labs, Dixon, CA) was driven via a syringe pump (PHD2000, Harvard Apparatus, Holliston, MA) through the phantom. The randomized sequence and image acquisition were controlled by a custom-designed LabVIEW program (National Instruments, Austin, TX). Only data from the most superficial capillary tube was analyzed. Each complete program of flow rates was repeated five times (i.e., $n=5$) for both epi-illumination and trans-illumination configurations.

For testing the effect of increasing depth on speckle signal attenuation, a single flow rate was applied to the phantom, and raw speckle images were captured at each capillary tube depth (0, 120, 360, and 720 μm). The image capture sequence at each depth setting was repeated five times for both epi-illumination and trans-illumination configurations.

2.3.

In Vivo Analysis of Deep Collateral Blood Flow in Trans-illuminated Mouse Hindlimb

All animal protocols were approved by the Institutional Animal Care and Use Committee at the University of Virginia and conformed to all regulations for animal use outlined in the American Heart Association Guidelines for the Use of Animals in Research. Male C57BL/6 mice ($n=10$) were purchased from Charles River Laboratory (Wilmington, MA) and housed in the animal facilities at the University of Virginia. To produce consistent flow increases in the collateral arteries in a superficial adductor muscle, the FAL scheme in Fig. 2 was chosen. This particular ligation pattern has been previously shown to produce consistent arteriogenesis in the collateral arteries of the gracilis adductor muscles^26–29 with minimal heterogeneity in the baseline collateral structure and with the predicted changes in flow direction from baseline. Mice, 9 to 13 weeks of age, were anesthetized (i.p. $120\mathrm{mg}/\mathrm{kg}$ ketamine, $12\mathrm{mg}/\mathrm{kg}$ xylazine, and $0.08\mathrm{mg}/\mathrm{kg}$ atropine), depilated, and prepped for aseptic surgery. On the left leg, an incision was made directly above and along the femoral artery, and a window of skin was dissected free and retracted directly above the superficial adductor muscles. The femoral artery was gently dissected from the femoral vein and nerve between the bifurcation of the superior epigastric artery and popliteal artery. Two 6.0 silk sutures were placed underneath the femoral artery, but not yet tied, immediately distal to the epigastric artery, which served as the origin of the muscular branch artery in all mice.

Fig. 2

Schematic of the primary gracilis adductor collateral pathways superimposed upon an image of a vascular cast of the gracilis muscle. Arrows indicate the predicted direction and magnitude of blood flow both pre- (yellow) and post- (white) femoral artery ligation (FAL) (denoted by “X”). Flow direction reverses from pre- to post-FAL in the collateral entry region nearest the saphenous artery (i.e., right side), but does not in the muscular branch entrance region (i.e., left side).

Upon placement of the ligature (untied), the mouse was transferred to the intravital microscopy station, and the exposed tissue was maintained under a constant drip of warmed Ringer’s solution ($137.9\mathrm{mmol}/\mathrm{L}$ NaCl, $4.7\mathrm{mmol}/\mathrm{L}$ KCl, $1.2\mathrm{mmol}/\mathrm{L}$ ${\mathrm{MgSO}}_4$, $1.9\mathrm{mmol}/\mathrm{L}$ ${\mathrm{CaCl}}_2$, $23\mathrm{mmol}/\mathrm{L}$ ${\mathrm{NaHCO}}_3$) that was prepared with 5% ${\mathrm{CO}}_2$ balance ${\mathrm{N}}_2$, and set to maintain 37°C fluid temperature. Mouse body temperature was maintained by a surgical heating pad. For imaging of gracilis collateral hemodynamics, we used the intravital microscopy setup shown in Fig. 2. Before baseline imaging, muscles were allowed 15 min to equilibrate while being superfused with warmed Ringer’s solution. After the equilibration period, a baseline (preligation) series of image acquisitions using epi-illumination and then trans-illumination laser orientation were taken at multiple fields of view to encompass the gracilis collateral arteries between the saphenous and the muscular branch arteries. Between each series of raw speckle images (epi- and trans-illumination), images of the collateral arteries were taken using brightfield epi-illumination through an objective-mounted fiber optic light guide to allow for lumenal diameter measurements (A08650, Schott Inc., Elmsford, NY). After baseline acquisition, the ligature was tightened, and stoppage of blood flow through the femoral artery was visually confirmed in all mice. After 30 min post-ligation, trans-illumination speckle and brightfield then epi-illumination and brightfield imaging sequences were repeated. Predicted baseline and post-ligation flow direction were confirmed in an identical preparation on a separate set of mice through fluorescent imaging of intravascularly injected fluorescent microspheres (FP-3070-2, Spherotech Inc., Forest Lake, IL).

Blood velocity analysis of laser speckle images was limited to well-defined collateral artery regions, such that bulk microvascular flow in the surrounding gracilis muscle flow was not considered. The assumption of Poiseuille flow was made, where wall shear rate is proportional to average blood velocity divided by tube diameter. This assumption is valid for most cases in the microcirculation.³⁰ Therefore, proportional wall shear rate [speckle shear stress (SSR)] was calculated using the NSI measurement of average blood velocity and vessel diameter ($D$) at each designated location [Eq. (4)]

3.1.

Influence of Illumination Source on Sensitivity as a Function of Depth in Tissue Phantoms

The results from the tissue phantom experiments are shown in Fig. 3. Both trans- and epi-illumination exhibited a strong linear relationship between NSI and absolute velocity [Fig. 3(a)]. However, the trans-illumination signal was not as significantly diminished with an increasing tissue depth, showing an $\sim 50\%$ reduction of the exponential decay of the speckle velocity signal with an increasing depth compared to epi-illumination [$1.20\times {10}^{-3}$ versus $2.70\times {10}^{-3}$, Fig. 3(b) at a constant velocity of $5\mathrm{mm}/\mathrm{s}$]. As such, trans-illumination maintained 40% of the initial superficial signal out to 720-μm deep within the tissue phantom versus 15% with epi-illumination.

Fig. 3

Trans-illumination maintains linear relationship between normalized speckle index (NSI) and increases sensitivity to deeper velocity signals in vitro. (a) Line graph of NSI for both epi-illumination (closed squares) and trans-illumination (open circles) configurations ($n=5$ independent trials) as a function of known blood velocity. Both configurations demonstrated similar and highly linear relationships. (b) Graph of speckle velocity (${\mathrm{NSI}}_x$), normalized to surface speckle velocity (${\mathrm{NSI}}_0$), as a function of depth for epi-illumination and trans-illumination and fitted with an exponential relationship (line). *$p<0.05$ between epi- and trans-illumination at the same depth.

3.2.

Trans-illumination LSF Microscopy Better Demonstrates Regional Variation of Hemodynamic Changes Along Collateral Arteries After FAL

A representative comparison of in vivo images created by epi-illumination and trans-illumination LSF is shown in Fig. 4. While epi-illumination LSF [Fig. 4(a)] was somewhat capable of visualizing the most superficial elements of the gracilis collateral network (e.g., the saphenous artery and upstream region of the muscular branch), it generally yielded much poorer quality images. Epi-illumination LSF images exhibited significantly more noise when compared to the trans-illumination images [Fig. 4(b)], with much of this noise being generated by superficial fat cells overlying the muscle [Fig. 4(a), white arrows), especially in the collateral entry region near the saphenous artery. Epi-illumination images were also susceptible to errors created by light being directly reflected off the tissue surface [Fig. 4(a), green arrows]. As seen in Fig. 4(b), these noise and error issues were entirely obviated by switching to trans-illumination.

Fig. 4

Comparison of LSF blood flow maps generated with epi- and trans-illumination. The change from low flow at the entrance regions with a point of convergence in the central regions pre-FAL to high flow continuously across the collateral arteries was apparent in all experiments ($n=10$). White box: markedly enhanced flow in the central collateral region. Yellow boxes: decreased post-FAL saphenous artery flow. Yellow arrows: enhanced flow in collateral entrance regions evident with trans-illumination. White arrows: noise comparison between epi- and trans-illumination. Cyan arrows: diminished post-FAL saphenous artery flow signal is cleaner with trans-illumination LSF.

Post-FAL, as expected, both the central [Fig. 4(b), white boxes] and entrance regions [Fig. 4(b), yellow arrows] of the gracilis collateral network experienced significant increases in the blood flow that were clearly evident with trans-illumination. Epi-illumination imaging was capable of capturing some of the characteristics of increased flow, but sensitivity within the collateral network was considerably poorer. The FAL also induced an expected drop in saphenous artery flow [Fig. 4(b), yellow boxes], which was evident with both epi- and trans-illumination. However, the pre-FAL epi-illumination signal was often saturated, and both the pre- and post-FAL saphenous flow signals were prone to considerable noise [Fig. 4(a), cyan arrows]. Finally, it is important to note that, while trans-illumination yielded better image quality, further improvements in spatial resolution could be realized by incorporating more sophisticated temporal laser speckle contrast analyses.^31,32 Limitations on camera control and frame rate precluded such analyses in the current study.

3.3.

Velocity and Shear Rate Changes Generated by FAL

Because there were no changes in depth of the collateral artery network within the gracilis muscle from pre- to 30 min post-FAL, NSI could be used to determine relative velocity changes [Fig. 5(a)]. When coupled with diameter measurements obtained from brightfield imaging, the velocity changes were used to calculate relative change in wall shear rate from pre- to 30 min post-FAL [Fig. 5(b)]. As expected, there was an increase in the blood flow and velocity along the gracilis collateral network. Interestingly, the percent increase in the velocity [$102.9\%\pm 9.5\%$ and $99.7\%\pm 11.9\%$ for the muscular branch (left) and saphenous artery (right) entrance regions, respectively, $p=0.84$] and shear stress ($102\%\pm 12\%$ and $97\%\pm 10\%$ for muscular branch and saphenous artery entrance regions, respectively, $p=0.73$) were not different ($n=10$ mice). The central region of the gracilis collateral network showed the largest relative change in velocity after FAL, going from negligible flow (minimal signal prevented accurate baseline measurement) pre-FAL to high velocity post-FAL [Fig. 4(b), white boxes]. The switch from convergent blood flow toward the central region of the gracilis muscle pre-FAL from both the saphenous and muscular branch arteries to high flow across the length of the collateral artery was present in all mice, matching the predicted hemodynamic patterns seen in Fig. 1.

Fig. 5

Blood velocity and wall shear stress changes induced by FAL. (a) Bar graphs of relative velocity signal (NSI) within the muscular branch (nonreversing) and saphenous artery (reversing) entrance regions. (b) Bar graphs of relative shear stress (SSR). *$p<0.05$ between pre- and post-FAL within the same region ($n=10$ mice).