This method can help answer key questions in the cardiac electrophysiologic field, about the structure of the extracellular space and cellular communication mechanisms. This technique's main advantages are that it has high throughput capabilities, and an improved spatial sampling frequency. Essentially, we can obtain measurements more quickly and with higher confidence than before.

Perinexus identification and program troubleshooting are difficult to learn without visual demonstration, because the perinexus is a relatively newly defined structure, and troubleshooting may not be intuitive for investigators who are unfamiliar with MATLAB. For grayscale images use numerical computing software to ensure that no pixel has an intensity greater than 255. Then, open up our paired image in image processing software, and zoom in on the perinexus.

It is critical to properly identify the perinexus. The first step is to identify a gap junction plaque, which has a striped appearance. Then, we look for two opposing membranes, in plane, out to about 200 nanometers.

Identify the gap junction plaque from it's pentylaminer structure. The beginning of the perinexus, is the point at which the two opposing cell membrane bilayers diverge. Display a scale bar in nanometers.

The starting and ending points of the perinexus outline will be 200 nanometers from the beginning of the perinexus. Next, select the freehand selection tool. Click and drag, or use a stylus to carefully trace up along the inner membrane of one cell, to the beginning of the perinexus, and back along the inner membrane of the second cell.

Close the selected area by releasing the mouse button or lifting the stylus. Then, set the line width to one pixel, and the foreground color to the highest intensity value for the image type. Such as white, for grayscale image.

Create a traced outline from the selection, and save the resulting image as a file type compatible with the analysis software, such as JPEG or TIFF. If needed, open the membrane separation distance analysis software, and change the save locations for the data and figures to be generated. Save the file and close it.

Then, run the program. Set the spatial derivative gradient threshold appropriately for center line identification. Set the scale and pixels per unit length.

Set the spatial lower and upper limits for the region of interest, with respect to the edge of the gap junction. Select either automatic or manual start point detection. Manual start point detection may be necessary for irregularly-shaped perinexi.

Then, open the image with the outlined perinexus. Click and drag to draw a box around the perinexus, excluding the closed end. Double-click inside the traced perinexus outline to crop the image and identify the center line.

If the start point is to be selected manually a crosshair cursor and center line will appear over the original image. Select a point outside the perinexus close to the desired starting point to continue the process. Once the process is finished, confirm that the center line stays within the perinexus, and properly intersects the start point.

Review the generated data and plot the perinexal width. If the center line was not properly identified and isolated, open the g-mag image array to determine an appropriate gradient threshold. We use the index tool to click around the center line, and the g-mag array to get an idea of which pixel we want the center line finding algorithm to select.

The gradient threshold should then be set just above the intensity value of those pixels. Select the index tool, and click on and around the center line to display the index value of the pixels that should be selected. Set the spatial derivative gradient threshold to just above the index value, and run the process again.

If the starting point was not detected correctly in the automated process, run the program again using manual start point detection. In this process, manual outlines are dilated in one pixel increments to count the number of pixels between the two edges. Each increment is added to a working image to generate a spatial derivative.

The original outline and the center line are discontinuities in its magnitude. After isolating the center line, it is refined by dilation, erosion, and a pathfinding algorithm. Perinexal width is presented as a function of distance from the start of the perinexus, or within a region of interest, and as an average of both of those functions.

As the perinexus orientation changed, over-or underestimations in perinexus width were observed, depending on the dilation pattern. Trigonometric correction produced results strongly correlated with images rotated to orient the perinexus horizontally. The algorithm was validated for different spatial resolutions, reference units, and image sizes.

Both experienced and inexperienced users traced the outline faster than they manually segmented the image. And the automated process had significantly greater spatial resolution. Experienced and inexperienced observers accurately identified significant differences in perinexal width between patients with and without preexisting Atrial Fibrillation.

These observers also accurately identified no significant difference between absolute gap junction widths in the same population. The perinexal and gap junction widths were consistent with previous reports. While attempting this procedure, remember to take your time with the outline, as even small deviations from the membrane can produce substantial errors at this scale.

Generally, individuals new to this method struggle because they're unfamiliar with the structure they're measuring, or they're unsure how to troubleshoot issues to start point identification or center line deviation. Seeing this image processing method performed is critical. As perinexus identification and algorithm troubleshooting are difficult to learn without it.

And we want to ensure that we're quantifying what we say we're measuring. This technique paves the way for researchers in the field of cardiac health, to explore higher resolution quantification of many levels of cardiac function. From nanoscale, extracellular spaces to clinical determination of ventricular efficiency or disfunction.

Though this method was demonstrated with electromicroscopy images, it can also be applied to other imaging techniques, such as cardiac echocardiography to more precisely quantify the mechanical function of the heart. This method may find applications in any imaging field because the program can measure the space between two defined edges, if the scale is set appropriately, and the edges are nearly parallel.

摘要