Mouse retinal vasculature is particularly interesting in understanding the mechanisms of vascular pattern formation. This protocol automatically measures the diameter of mouse retinal vessels from fluorescent angiography fundus images at a fixed distance from the optic disk.
It is important to study the development of retinal vasculature in retinopathies in which abnormal vessel growth can ultimately lead to vision loss. Mutations in the microphthalmia-associated transcription factor (Mitf) gene show hypopigmentation, microphthalmia, retinal degeneration, and in some cases, blindness. In vivo imaging of the mouse retina by noninvasive means is vital for eye research. However, given its small size, mouse fundus imaging is difficult and might require specialized tools, maintenance, and training. In this study, we have developed a unique software enabling analysis of the retinal vessel diameter in mice with an automated program written in MATLAB. Fundus photographs were obtained with a commercial fundus camera system following an intraperitoneal injection of a fluorescein salt solution. Images were altered to enhance contrast, and the MATLAB program permitted extracting the mean vascular diameter automatically at a predefined distance from the optic disk. The vascular changes were examined in wild-type mice and mice with various mutations in the Mitf gene by analyzing the retinal vessel diameter. The custom-written MATLAB program developed here is practical, easy to use, and allows researchers to analyze the mean diameter and mean total diameter, as well as the number of vessels from the mouse retinal vasculature, conveniently and reliably.
Possibly the most researched vascular bed in the body is the retinal vasculature. With ever-improving technical sophistication, retinal vasculature is easily photographed in living patients and used in many research fields1. Additionally, the mouse retinal vasculature during development has proven to be a very effective model system for research into the fundamental biology of vascular growth. The primary purpose of the retinal vasculature is to provide the inner portion of the retina with metabolic support through a laminar capillary meshwork that permeates the neural tissue2. Nevertheless, the condition of the retina, and consequently any dysfunction or atrophy, can have significant effects on both the bifurcations of the retinal vasculature and the diameter of arteries, demonstrating an interplay between the retinal cells and the vasculature3,4. It is known that numerous eye conditions, including retinopathy of prematurity (ROP), diabetic retinopathy (DR), age-related macular degeneration (AMD), glaucoma, and corneal neovascularization, can result in abnormal ocular angiogenesis5. In the case of the retinal vasculature, mouse models of retinal degeneration often exhibit changes that are comparable to those seen in human vascular diseases6,7. The Myc supergene family of fundamental helix-loop-helix-zipper transcription factors includes the microphthalmia-associated transcription factor (Mitf) gene expressed in the retinal pigment epithelium (RPE)8,9,10.
Numerous organs, including the eye, ear, immune system, central nervous system, kidney, bone, and skin, have been demonstrated to be regulated by Mitf9,11,12,13. We have discovered that the structure and function of the RPE are affected in mice carrying various mutations in the Mitf gene, resulting in some cases of retinal degeneration and, ultimately, vision loss10. Recently, it has been shown that the number of vessels and vessel diameter differ significantly between Mitf mutant and wild-type mice14. Researchers and physicians can now precisely quantify the retinal vasculature in vivo due to retinal imaging developments. Since the 1800s, researchers and physicians have taken advantage of the benefit of visualizing the retinal vasculature, and fluorescein angiography (FA) has shown both retinal blood flow and degradation of the blood-retinal barrier15.
This article demonstrates how to analyze the retinal vessel diameter from mouse FA images with a custom-written code in MATLAB software.
All experiments were approved by the Icelandic Food and Veterinary Authority (MAST license No. 2108002). All animal studies were conducted according to the Association for Research in Vision and Ophthalmology (ARVO) Statement for the Use of Animals in Ophthalmic and Vision Research. Male and female C57BL/6J and Mitfmi-vga9/+ mice were used in this study. C57BL/6J mice (n = 7) were used as a control. The wild types were commercially obtained (see Table of Materials), but all mutant mice (n = 7) were bred and raised at animal facilities in the Biomedical Center at the University of Iceland. In the present study, 3-month-old animals were used; however, the protocol applies even to 1 month and older animals.
1. Experimental preparation
2. In vivo imaging of retinal vasculature using a rodent retinal imaging system
3. Analysis of the retinal vessel's diameter
Figure 1 shows the process used to analyze the retinal vasculature, which is applied to mouse FFA images from all the tested mice. A radius that is twice as large as the optic disc is used to measure the intensity of pixels in a circular, clockwise direction from the optic disc's center. It marks pixels with a start or end point when it comes across points above and below a user-specified threshold, respectively. This is repeated 30 times, each time going a little bit further away from the center of the optic disk. FFA images are taken 5 min after fluorescein injection (Figure 1A and Figure 1B). The analysis software then calculates the shortest route between each of the 30 relevant end points and each of the start points. The outcome of this processing is a figure with the points and the measurements represented by a white line between them; examples of such images from a control and Mitf mutant mouse are shown in Figure 1A and Figure 1B, respectively. However,the black arrows shown in Figure 1B are minor errors that may occur, but the software does not read these errors as retinal vessels. Moreover, the program calculates the mean vessel diameter of each retinal vessel in pixels based on the 30 end points determined on the vessel, starting from the superior central vessel above the optic disk (as vessel 1) and then clockwise around the disk in numerical order. The software then presents the mean vessel diameter in a graph as a function of vessel number; examples of these graphs from a wild-type and mutant mouse are displayed in Figure 2A and Figure 2B, respectively. Along with the measurement values for each vessel, it also offers an Excel document with the mean, median, and standard deviation computations of vessel diameter in pixels. These tables are obtained for all the mice investigated in this study (Table 1 and Table 2) from the software; therefore, the values are not copied into another program. The value "N" in Table 1 and Table 2 corresponds to the number of points taken from each vessel and calculated by the program. When this value is less than or equal to 5, the software reads it as not applicable (N/A), and therefore it does not correspond to a retinal vessel.
Figure 1: Fluorescein angiography images. The images from a control (A) and Mitfmi-vga9/+ mutant (B) mouse used in this study. The red dots correspond to the center and the edge of the optic disk. The white lines on the vessels show where measurements of the shortest distance between points are taken. The black arrows indicate possible errors in the code. Please click here to view a larger version of this figure.
Figure 2: The mean vessel diameter as a function of vessel number. Mean vessel diameter (in pixels ± standard deviation [SD]) of the main retinal vessels in a fundus image from a wild-type (A) and Mitfmi-vga9/+ mutant (B) mouse. The abscissa in both panels indicates the vessel number of each vessel, counting from the superior central vessel above the optic disk as vessel number one, and then clockwise from that vessel. Please click here to view a larger version of this figure.
Table 1: Mean retinal vessel diameter (in pixels) in the fundus of a wild-type mouse. The values in the table are obtained from the analysis software. Each line represents data from one vessel, and the first line is the vessel in the superior central part of the fundus above the optic disk. The vessels are then numbered in a clockwise direction from that vessel. Please click here to download this Table.
Table 2: Mean retinal vessel diameter (in pixels) in the fundus of a Mitfmi-vga9/+ mutant mouse. The same procedure is used as in Table 1. The values in the table are obtained from the analysis software. Each line represents data from one vessel, and the first line is the vessel in the superior central part of the fundus above the optic disk. The vessels are then numbered in a clockwise direction from that vessel. Please click here to download this Table.
Supplementary File: "fundusDiameter.m" code for determining the retinal vessel's diameter. Please click here to download this File.
The present article is the first to present a method to analyze retinal vessel diameter and retinal vasculature from mouse FA images. Since only fundus imaging was utilized to capture images of the retinal vasculature, the method has several drawbacks, one of which is that one can only infer alterations in the superficial layers of that the retinal vasculature in the mice examined in this study; any differences in the deeper layers are yet unknown.
A unique optical coherence tomography angiography (OCTA) image analysis method using automatic vessel tracing and vessel diameter has been presented16. However, this method requires a gradient-guided minimum radial distance (MRD) measurement, which is the basis of an automated framework for the quantitative characterization of the diameter of blood arteries, including individual capillaries. Furthermore, the sizes of tiny capillaries are probably overstated to some extent, since the OCT system has limited lateral resolution16. This method requires FA, which can capture a much wider area of the retinal and choroidal vasculature17. The Dynamic Vessel Analyzer (DVA) prototype can be used to analyze retinal vessels in mice18. Nevertheless, this device uses flicker light impulses applied at predetermined frequencies to enable dynamic vessel examination as a function of time. As shown by postmortem analysis, when comparing flicker-exposed with flicker-naïve retinae, visual stimulation during the recording method may put stress on the investigated retina18. This method allows the analysis of the retinal vasculature and retinal vessel diameter in both control and Mitf mutant animals under anesthesia without previous visual stimulation through the use of fluorescein salt to enhance contrast. However, other mouse models with different mutations, as well as albino mice, have not been investigated with the current method. Further investigation is needed to explore and analyze vessel diameter in other strains of mice.
Since the fundus camera is equipped with a rat objective lens, we believe that this method could be applied to rats. As different anesthetics have been demonstrated to impact functional retinal blood flow in the rat eye, they are likely to also impact retinal vascular diameter19; as a result, it is crucial to emphasize that choosing an anesthetic requires careful thought. A combination of ketamine/xylazine was used in this study, and it is possible that this had some impact on how the retinal vascular sizes were measured. An alternative anesthesia method could be isofluorane inhalation via a mask, frequently used for similar procedures in mice.
It is crucial to time the acquisition and analysis of retinal vascular images so that the diameter measurements can be made 5 to 10 min following intraperitoneal fluorescein injection. In fact, defining the ideal period of picture acquisition as carried out in this study in a single example may be advised for each laboratory and study setup individually. The program intensity threshold was set to 80% on every image under analysis. The vessels were put through the MATLAB program after being carefully chosen. The reproducibility of the results was evaluated as a part of the MATLAB program’s validation. It is also possible that the fundi of some of the Mitf mutant mice had background levels that are greater than the 80% cutoff, but the software would have recognized this, and is therefore equally likely to be a small error. Branching is a variable that affects the analysis of the retinal vasculature and must be properly considered while using the program to carry out the measurements. The process of determining vascular diameter must be entirely automatic in the future, but this has proven to be a challenging task thus far due to the variations in fluorescence intensities, potential vascular leakage, and other variations in the retinal vasculature between our Mitf mutant mouse model.
The authors have nothing to disclose.
This work was supported by a Postdoctoral Fellowship grant from the Icelandic Research Fund (217796-052) (A.G.L.) and the Helga Jónsdóttir and Sigurlidi Kristjánsson Memorial Fund (A.G.L and T.E.). The authors thank Prof. Eiríkur Steingrímsson for providing the mice.
1% Tropicamide (Mydriacyl) | Alcon Inc Laboratories | Mydriatic agent | |
2% Methocel | OmniVision Eye Care | Hydroxypropryl methylcellulose gel | |
C57BL/6J | Jackson Laboratory | 000664 | Wild type mice |
Chanazine 2% (xylazine) | Chanelle Animal Health UK | BN I21322/I | Anesthesia IP |
Excel for Microsoft 365 | Microsoft Inc | Software package | |
Fluorescein sodium salt | Sigma-Aldrich | 28803-100G | Fluorescent angiography |
Matlab 8.0 | The MathWorks, Inc. | Software package | |
Micron IV rodent fundus camera | Phoenix-Micron | 40-2200 | Fundus photography |
Phenylephrine 10% w/v | Bausch & Lomb | Mydriatic agent | |
Phosphate Buffered Saline – 100 tablets | Gibco | 18912-014 | Dilution |
Sigmaplot 13 | Jandel Scientific Software | Software package | |
S-Ketamine, 25 mg/mL | Pfizer Inc. | PAA104470 | Anesthesia IP |