This study demonstrates the surgical preparation of the rat cremaster muscle for the visualization of the in vivo cell-free layer. Considerable factors affecting the accuracy of the cell-free layer width measurement are discussed in this study.
The cell-free layer is defined as the parietal plasma layer in the microvessel flow, which is devoid of red blood cells. The measurement of the in vivo cell-free layer width and its spatiotemporal variations can provide a comprehensive understanding of hemodynamics in microcirculation. In this study, we used an intravital microscopic system coupled with a high-speed video camera to quantify the cell-free layer widths in arterioles in vivo. The cremaster muscle of Sprague-Dawley rats was surgically exteriorized to visualize the blood flow. A custom-built imaging script was also developed to automate the image processing and analysis of the cell-free layer width. This approach enables the quantification of spatiotemporal variations more consistently than previous manual measurements. The accuracy of the measurement, however, partly depends on the use of a blue filter and the selection of an appropriate thresholding algorithm. Specifically, we evaluated the contrast and quality of images acquired with and without the use of a blue filter. In addition, we compared five different image histogram-based thresholding algorithms (Otsu, minimum, intermode, iterative selection, and fuzzy entropic thresholding) and illustrated the differences in their determination of the cell-free layer width.
In vivo animal studies are instrumental to basic science for understanding human physiology and pathology. In particular, in vivo microhemodynamic studies can elucidate the potential impairment of microcirculatory functions altered by abnormal rheological conditions of blood. A number of previous microhemodynamic studies1 have used the rat cremaster muscle model for visualizing microvascular blood flow. The cremaster muscle is a thin layer of striated muscle surrounding the testes. Thus, the blood flow in the muscle can be visualized with a trans-illumination microscope by means of surgical exposure. This enables us to acquire the in vivo blood flow images without the use of any fluorescence or contrast agents. In addition, the entire blood perfusion of the muscle network can be controlled by reducing the upstream blood flow with abdominal aorta occlusion2. Owing to these advantages, the cremaster muscle model has been widely used to investigate the formation of cell-free layer (CFL) in microvessels1,3.
The CFL width is a prominent hemodynamic parameter in microcirculation, which has been of great interest for its important roles in regulating microcirculatory functions. The CFL is formed by the shear-induced transverse inward migration of red blood cells (RBCs) towards the flow center4. Consequently, this migration leads to the depletion of RBCs near the vessel walls, eventually resulting in a cell-free plasma layer. Accordingly, the parietal CFL naturally becomes a diffusion barrier to oxygen (O2) delivery from the RBC core to the tissues, and to the scavenging of nitric oxide (NO) by the RBCs5,6. In addition, the production of NO can also be modulated by the dynamic variations of the CFL width7,8. Therefore, the roles of the CFL in both gas transport and the regulation of homeostasis in microcirculation need to be fully ascertained to better understand blood flow in microcirculation. Recent studies have focused on bridging the hemodynamics and gas transport functions of the CFL in the microcirculation9-12. Furthermore, a separate set of studies has also investigated how the pathological elevation in RBC aggregation modulates CFL formation and its effect on O2 and NO bioavailability in tissues13,14.
The roles of the CFL become more significant in microcirculation where the relative size of the CFL width to the vessel diameter is prominent. This necessitates an effective approach of quantifying the CFL in in vivo blood flow. Particularly, image acquisition and image analysis are the two key components determining the accuracy of CFL width measurement. Successful visualization of tissue blood flow should be preceded by an appropriate surgical preparation of the animal model. Additionally, a proper image analysis technique is needed to overcome the limitations of conventional manual measurements that are mostly induced by human errors15,16. With advancements in optical instrumentation and computing power for digital image processing, it is now possible to achieve a more accurate and consistent measurement of the CFL width17-19. Nonetheless, the accuracy of these measurements, being image-based, still ultimately depends on the quality of the images.
Therefore, this study explores the factors influencing the measurement of the in vivo CFL width. We focused particularly on demonstrating the surgical preparation and digital image analysis for measurements of the CFL width in arterioles of the rat cremaster muscle.
This study is in accordance with the National University of Singapore Institutional Animal Care and Use Committee (approved protocol no. R15-0225).
1. Surgical Preparation of the Animal Model
2. Image Analysis
The visualization of the CFL in vivo is largely dependent on the surgical preparations of the animal. Excessive blood loss or extended surgery duration may subject the animal to shock and blood flow aberrations. Maintenance of tissue temperature using a heating pad as well as a customized platform during the surgery and experiment is also crucial for maintaining the physiological conditions of the rat. By using a 100 W halogen lamp in the microscope system, no discernible tissue damage was observed even at the end of the experiment.
Figure 2A shows a typical RBC flow through an unbranched arteriole in the rat cremaster muscle, where the CFL can be observed between the RBC core and the inner vessel wall (Figure 2C). A good contrast between these components during the experiment is critical for ensuring the accuracy of CFL width measurements. The initial phase of the image analysis involves the detection of the inner vessel wall. By acquiring the light intensity profile along the analysis line perpendicular to the vessel, the location is approximated at the peak that transits from dark to light over two pixels (Figure 2B).
As RBCs and CFL possess different light transmittance, the difference in gray levels can be subdivided into two classes (binary image). However, the identification of an accurate threshold value between the two peaks in the image histogram may be restricted by poor image quality and contrast (Figure 3A). To improve the contrast between the RBCs and CFL, a blue filter can be used (Figure 3B). This is also evident in Figure 4, in which the boundaries of the RBC core can be more accurately identified with the use of a blue filter. Furthermore, the selection of thresholding algorithm20-23 can also influence the measurement of the CFL width (Figure 4). It is apparent in Figure 4A that different thresholding algorithms resulted in different RBC core boundary identified, which could in turn lead to erroneous CFL measurements. To better illustrate the influence of the thresholding algorithm on the CFL width measurement in Figure 4B, the spatial profiles for the CFL widths obtained using different thresholding algorithms are shown in Figure 5 and summarized in Table 1.
Figure 1: Intravital Microscopic System and Cremaster Muscle Preparation. A: Surgically exteriorized rat cremaster muscle. B: Customized platform with heating elements for placing the cremaster muscle and maintaining its temperature at 35 °C. C: Microscopic system with customized animal stage and high-speed camera for the visualization of microcirculatory blood flows in the cremaster muscle. D: Negative feedback temperature controller and power supply. E: Physiological data-acquisition system for continuous pressure monitoring. Please click here to view a larger version of this figure.
Figure 2: Image Processing for Determination of Vessel Wall Position and CFL Width. A: Typical grayscale image of RBC flow in an arteriole (vessel diameter = 52 µm). B: Light intensity profile along the analysis line (solid line in panel A). C: Representative result of CFL measurement along the vessel. The solid and dashed arrows indicate the inner vessel wall and outer edge of the RBC core, respectively. (LWB & RWB: left and right vessel wall boundary, LCB & RCB: left and right RBC core boundary) Please click here to view a larger version of this figure.
Figure 3: Image Contrast Enhancement with an Optical Blue Filter. Image histogram of the grayscale images obtained without (A) and with blue filter (B). Please click here to view a larger version of this figure.
Figure 4: RBC Core Width Determined using Five Different Thresholding Algorithms. Boundaries of RBC core and vessel wall superimposed on grayscale images in Figure 3. (Top row (A): without blue filter, bottom row (B): with blue filter) using (from left to right) the Otsu's method, minimum method, intermode method, iterative selection method (Isodata) and fuzzy entropic thresholding (Shanbhag). The solid and dashed lines indicate the inner vessel wall and outer edge of the RBC core, respectively. Please click here to view a larger version of this figure.
Figure 5: Spatial Variation of CFL Width. CFL widths corresponding to Figure 4B along the left (A) and right (B) vessel walls, respectively. (D: distance in vessel diameter) Please click here to view a larger version of this figure.
Table 1: Threshold Values and CFL Width Data in Figure 5. * p < 0.001: significant difference from Otsu's method. † p < 0.001: significant difference from left. Statistical analyses were performed using two-tailed unpaired t-tests.
The measurement of CFL width is essential for a better understanding of the hemodynamics in the microcirculation. In particular, the measurement of CFL widths has been performed in mesenteric6, spinotrapezius24 and cerebral25 microcirculations. Conventional measurement of in vivo CFL widths was restricted to estimations by manual inspection of the recorded video frames. The manual measurements required the averaging of several successive video frames before visually identifying the boundaries of the RBC core and vessel walls15,16. In another study, fluorescein isothiocyanate (FITC)-labelled RBCs and rhodamine-B isothiocynate (RITC) labelled plasma were used to determine the mean CFL widths in cat cerebral microvessels25. These previous measurement methods are very time consuming and require additional steps for the fluorescent labelling, which limits the spatial and temporal resolution of the CFL width measurement. In contrast, by coupling high-speed camera recordings to an effective image segmentation and analysis, the technique demonstrated here permits the quantification of spatiotemporal variations of the CFL with a spatial resolution (0.42 µm) of an order smaller than the size of a RBC and a temporal resolution of 1/3,000 sec.
Proper surgical preparation of the cremaster muscle is critical in determining the accuracy of the CFL width measurements. In particular, thorough removal of adjacent connective tissues is essential to ensure a good focus of the arterioles in the cremaster muscle. In addition, the temporal and spatial resolution of the measurement is dependent on the microscope and camera specifications. While a higher magnification objective may enhance the spatial resolution, it reduces the field of view, which in turn limits the obtainable vessel length for quantifying the spatial variation of the CFL width. Therefore, the microscopic configurations can be modified according to the specific application of the technique.
Image segmentation is another important factor for the accuracy of the CFL width measurement. Amongst the various techniques developed, image thresholding based on gray level histogram provides a simple and effective approach for image segmentation and analysis. Accordingly, foreground objects are extracted from the background based on the difference in their gray levels. In the ideal case, the image histogram will be bimodal and a threshold value at the bottom of the valley is trivial. However, in vivo experimental images do not always exhibit such grayscale level profiles. Our results have shown how the image quality and contrast can influence the image segmentation process. The use of an optical blue filter significantly enhanced the contrast between the RBCs and the plasma in an arteriole (Figure 3), and it seems to be essential when applying the histogram-based thresholding for the CFL width measurement regardless of the algorithms (Figure 4). This results in a distinct bimodal image histogram, which allows one to identify the threshold value effectively. However, it should be noted that even with a bimodal histogram obtained from the in vivo images, an extremely unequal variance of two peaks (local maxima) and a wide valley (local minimum) of the histogram can still influence the threshold selection (Table 1). Therefore, the selection of an appropriate thresholding algorithm needs to be examined based on the image quality and users have to consider the limitations of each thresholding algorithm for the best suitability in quantifying CFL widths.
As the CFL widths are largely dependent on the flow conditions, continuous arterial pressure measurement throughout the course of the experiment is essential. In order to determine the local flow conditions, the pseudoshear rate of the blood flow can be computed by measuring the mean flow velocity in the blood vessel5.
In summary, the protocols for the surgical preparation of a rat cremaster muscle and quantitative image analysis described here have been utilized to acquire quantitative information on the dynamic variation of the CFL widths in vivo. The primary challenges in ensuring the accuracy of the CFL width measurements include proper surgical preparation of the muscle and image segmentation, both of which have been addressed above. This technique can be readily adapted to other microcirculatory studies to investigate the hemorheological and hemodynamic aberrations in various physiological and pathological conditions. Consequently, these findings contribute to the future development of microvascular therapeutic approaches and clinical intervention.
The authors have nothing to disclose.
This work was supported by National Medical Research Council (NMRC)/Cooperative Basic Research Grant (CBRG)/0078/2014.
Intravital microscope | Olympus | BX51WI | Equipment |
High speed camera | Photron | 1024PCI | Equipment |
Blue filter | HOYA | B390 | Equipment |
Pressure sensor &biopac system | Biopac system | TSD104A, MP100 | Equipment |
Temperature controller | Shimaden | SR 1 | Equipment |
Plasma Lyte A | Baxter | NDC:0338-0221 | Warm in 37 °C water bath before use |
Saline 0.9% | Braun | ||
Heparin (5000 IU/ml) | LEO | ||
PE-10 polyethylene tube | Becton Dickinson | 427400 | .024" OD X .011" ID |
PE-50 polyethene tube | Becton Dickinson | 427411 | .038" OD X .023" ID |
PE-205 polyethene tube | Becton Dickinson | 427446 | .082" OD X .062" ID |
2-0 non-absorbable silk suture | Deknatel | 113-S | |
5-0 non-absorbable silk suture | Deknatel | 106-S | |
Water circulating heating pad | Gaymar | ||
Water bath | Fisher Scientific | Isotemp 205 | Equipment |
Sterile Cotton Gauze | Fisher Scientific | 22-415-468 | |
Cotton-tipped applicators | Fisher Scientific | 23-400-124 | |
Dumont Forceps | Kent Scientific | INS14188 | Surgical instrument |
Micro Dissecting forceps | Kent Scientific | INS15915 | Surgical instrument |
Iris forceps 1×2 teeth | Kent Scientific | INS15917 | Surgical instrument |
Vessel cannulation forceps | Kent Scientific | INS500377 | Surgical instrument |
Micro scissor | Kent Scientific | INS14177 | Surgical instrument |
Iris scissor | Kent Scientific | INS14225 | Surgical instrument |
Vessel clip | Kent Scientific | INS14120 | Surgical instrument |
Gemini cautery system | Braintree Scientific | GEM 5917 | Surgical instrument |