Intravital imaging-based analysis tools for vessel identification and assessment of concurrent dynamic vascular events

The vasculature undergoes changes in diameter, permeability and blood flow in response to specific stimuli. The dynamics and interdependence of these responses in different vessels are largely unknown. Here we report a non-invasive technique to study dynamic events in different vessel categories by multi-photon microscopy and an image analysis tool, RVDM (relative velocity, direction, and morphology) allowing the identification of vessel categories by their red blood cell (RBC) parameters. Moreover, Claudin5 promoter-driven green fluorescent protein (GFP) expression is used to distinguish capillary subtypes. Intradermal injection of vascular endothelial growth factor A (VEGFA) is shown to induce leakage of circulating dextran, with vessel-type-dependent kinetics, from capillaries and venules devoid of GFP expression. VEGFA-induced leakage in capillaries coincides with vessel dilation and reduced flow velocity. Thus, intravital imaging of non-invasive stimulation combined with RVDM analysis allows for recording and quantification of very rapid events in the vasculature.

fast flow (dark grey ellipse; long green line), a larger distance is travelled by the scanning laser before a lower tangent is made with the moving RBC compared to that captured under slower flow (light grey ellipse; short green dashed line), resulting in greater image distortion (compare long and short red arrows). h. Schematic representation of the method used to find Δx and Δy, i.e. the distance travelled within the laser scan field by a moving RBC along the x-and y-axes respectively. Calculated lengths (Xm, Ym; black arrows) and measured lengths (Xs and Ys; green arrows) from distorted RBC images in sXYT respectively, were used to calculate Δx and Δy (red arrows).
Supplementary Supplementary note 1

RVDM method Walk-through and trouble-shooting guide
Calculation of actual RBC size using fXYT imaging Actual RBC dimensions 2a and 2b whilst under flow in C57BL/6J mice are provided in the analysis spreadsheet. Be aware that in certain situations, RBC size and shape may differ from our measured parameters. For example, species or strain variation may exist or may be affected in pathological settings, such as during obese hyperglycaemia 1 . To ensure accurate measurement of flow velocity, it may be necessary to re-calibrate the spreadsheet by measuring actual RBC diameter (2a and 2b) using fXYT imaging. For re-calibration, an acquisition rate under 20 msec / frame is recommended, note however that faster acquisition rates may be required for vessels with very fast flow speeds, to prevent RBC image distortion. From these fast acquisition movies, measurement of RBC length parallel and orthogonal to flow will give 2a and 2b respectively (See Supplementary Figure 5). These measurements should be repeated for RBCs under different flow velocities with n=30 or higher for accurate measurements. Lengths 2a and 2b can then be entered into the spreadsheet in the green highlighted boxes (see Supplementary Figure 6). In-frame analysis 1. Ensure that bounding rectangle is selected to be measured within ImageJ (Fiji) (Analyze → Set Measurements → Bounding rectangle) and that in-frame analysis is selected within the spreadsheet. 2. From acquired movies of flowing RBCs choose an RBC image with a high S/N ratio and clear edges. 3. Draw a line between the upper and lower tangent points formed between the laser scan line and RBC (i.e. c → c' (not a → a') (see Supplementary Figures  2 and 8). 4. Measure the line and insert the line width, height and length into the spreadsheet under their respective headings. 5. Repeat for other RBCs within the same vessel (recommended n=20 or higher).

Frame analysis
1. Ensure that frame analysis is selected on the spreadsheet. 2. Draw a line with ImageJ linking the same point on an RBC between one frame and the next (see Supplementary Figure 9). 3. Measure the line and insert the length into the spreadsheet under their respective headings. 4. Repeat for other RBCs within the same vessel (i.e vessel sections between intersections and bifurcations). Recommended n=10 or higher.

Scan line direction
Flow direction .between one frame and the next.

Results
The spreadsheet will calculate the velocity of RBCs from the entered information. The spreadsheet will also determine the reliability of individual data points and reject some based on comparison between Vx and Vy, the flow angle and their deviation from the mean. The spreadsheet will then return the average RBC velocity, the standard deviation and the number of values that have contributed to these values in the upper left corner.
Troubleshooting Q: In which direction is the blood flowing? Determining flow direction is imperative to calculation of velocity but can be difficult. Direction of flow can however be determined from the shape of the RBC image.
For example, if the line c → c' runs from left to right then then flow is running from left to right and vice versa. See an example in Supplementary Figure 8.
In addition, in cases were flow has a vertical orientation, RBC images will elongate or shrink compared to their actual size when flow is with scan direction (i.e. top to bottom) and against scan direction (i.e. bottom to top), respectively.
Q: How do I know if an RBC image is made of 1 or 2 RBCs? It is sometimes not clear whether RBC images acquired by sXYT represent single or multi RBC's in vessels with high velocity and high hematocrit.
When measuring RBC dimension c → c', single RBC shapes with a clear top and bottom border should be chosen (green arrows, Supplementary Figure 10). The lines indicated by yellow arrows could also be used but might be rejected by the RVDM error checks. The lines indicated by orange arrows shouldn't be chosen for analysis and any measurements taken from these would be rejected by the RVDM error checks.
Supplementary Figure 10. Example images illustrating the quality of RBC image, captured by sXYT, which is required for analysis using RVDM. Green arrow, acceptable RBC; yellow arrow, may be tolerated; orange arrow, rejected.
Q: No or very small RBC images appear in vessels; why? Situations where capillaries have no RBCs running through them can occur when flow is preferentially running through other routes. If this happens then imaging for a longer period may allow for a sufficient number of RBCs to pass. Alternatively, smaller fluorescent particles (no less than 1 µm) may be injected into the blood stream and used to calculate flow velocity.
This situation may also arise when flow velocity is too fast for the scanning parameters, resulting in the laser being able to capture an insufficient portion of the RBC before it passes. In these situations, parameters should be changed to enhance scanning speed.
Q: What do I do if the data I insert into the analysis spreadsheet produces no results? As the formulas used to calculate velocity reject those data deemed to be unreliable, for example because an RBC image is actually 2 adjacent RBCs or the S/N ratio is poor, situations may arise where the spreadsheet produces no final result or very few n's. This may be more apparent in vessels where the angle of blood flow θ is near 0, 90, 180 or 270˚ where measurement of Ys or Xs is less reliable. In such situations, measurement of more RBCs may be necessary or measurements may be taken from a section of the vessel where θ is farther from 0, 90, 180 or 270˚. However, in situations where the S/N ratio is too poor new experimental acquisition may be necessary.
Q: How do I decide whether to use the frame or the in-frame method? This is dependent on the scan speed that is used, however as the benefit of RVDM over for example fXYT is its compatibility with acquisition of wide fields of view using slower scan speeds, the in-frame method would most commonly be used. The frame method may likely only be used in those situations where blood flow is very slow and a single RBC can be tracked from one frame to the next. For example, for the tissue and scan parameters used in this study approximately 85% of vessels required inframe analysis.
Q: How much time is required to acquire and analyze a field of vessels? This is dependent on the size of the field, the number of vessels present and the experience of the user. However, acquisition of a field of vessels should take minutes, to ensure capture of a high-resolution 3-dimensional image. The analysis of flow velocity would then be expected to take 2-4 minutes per vessel for an experienced operator.
Q: Do I need to acquire RBC images at different scan angles and speeds?
In order to carry out accurate velocity quantification using RVDM, additional scan angles should not be necessary. On the other hand, additional scan speeds could be useful for vascular beds that have a very broad range of flow speeds. In this case we believe that only two scan speeds would be necessary, one at a normal scan speed as we have used in this study and one faster to allow analysis of very fast flowing arteries.