Three-dimensional (3D) reconstruction of the AV junction
All sections collected from 80 groups (320 sections) from one rat were
used to reconstruct a 3D computational model of the anatomy of the AVN.
3D-DOCTOR (version 4.0; Able Software Corp.) and Adobe Photoshop were
used to analyze the images and construct a 3D anatomical model of the
AVN. There were four steps in this process. Step 1:
Segmentation. Various tissue types could be identified in Masson’s
trichrome-stained sections: myocytes were stained purple, and connective
tissues were stained blue. Further cell types were identified by merging
adjacent immunofluorescence-labeled images with images from similarly
grouped Masson’s trichrome-stained sections using Adobe Photoshop, and
these combinations included the following: (i) Cx43-negative,
Cx40-negative, HCN4-positive and strongly Cx45-positive nodal tissues;
(ii) Cx43-negative, Cx40-positive, HCN4-positive and strongly
Cx45-positive His bundle tissues; and (iii) Cx43-positive ventricular
and atrial muscles. Then, Adobe Photoshop was used to outline various
cell types (carried out manually) on images obtained from Masson’s
trichrome-stained sections. A unique color was used to highlight each
cell type. Finally, multicolor images were converted into grayscale
images in preparation for 3D reconstruction. The result was a
two-dimensional model of sections showing the distribution of myocytes,
connective tissues and different cell types. Step 2: Alignment.Sections were aligned by some combination of translation, rotation and
scaling, but the shape of the sections was left unchanged. The alignment
of sections is crucial in developing a model. Tracts of the nodal
tissue, tricuspid valve and bottom edge of the preparation were the main
landmarks used to align the sections. Each image was aligned by
3D-DOCTOR and saved. The saved aligned images were used in preparation
for the next step of 3D reconstruction. Step 3: Calibration .
The voxel size (image resolution) must be provided to obtain a 3D image
so that the 3D rendering has the correct scale in all three dimensions,
enabling it to incorporate the reported measurements. For the voxel
resolution, the X and Y values are equivalent to the size of one pixel
in a slice, while the Z value is the slice thickness and distance
between the centers of two neighboring slices in 3D space. A group at
the same level of the tissues was created as an assumption; thus, the
distance between two prepared grayscale images was 24/28 μm (6/7 μm*4).
Volume correlation measurements were based on parameters obtained from
the image and scale plate provided by the camera (1 μm = 1.17 unit). As
a result, Z values (28.08/32.76 units) were used as the thickness
between the two groups of sections, which is an option for 3D
reconstruction. Step 4: Model visualization. After creating an
interactive segment, surface rendering was performed using 3D-DOCTOR.
Although only two preparations were used to generate an anatomical
model, we confirmed these anatomical findings in four other
preparations.