Reconstruction of 3-Dimensional Histology Volume and its Application to Study Mouse Mammary Glands

The overall goal of the following video is to present an image registration approach for three dimensional histology, volume reconstruction to study changes in mammary cellular macro structure. To do this, mammary glands are excised from mice three days after the onset of lactation in wild type and insulin-like growth factor binding protein seven, IGF BP seven null mice. The glands are then processed and embedded in paraffin blocks for sectioning.

Next optical block face images of the tissue blocks are captured and the tissue block is sectioned. Images of the sections are captured and using custom software, each section is aligned with its corresponding block face image to reconstruct histology volume of the mammary gland. The resulting data highlight differences in size in mutant and wild type glands.

The main advantage of this technique over existing methods like three dimensional confocal twofold microscopy, is that it provides information over a much greater spatial extent. This method can also be used to investigate and validate novel techniques and algorithms in volumetric medical imaging and therapies, as well as creating 3D high resolution analysis of different organs. This reconstructed histology volume portraits a reliable representation of the process specimen with no propagated registration error.

We first had the idea for this method when we noticed that IGF P seven null mice have smaller litter sizes and are unable to sustain multiple large litters, so we decided to investigate the structure of their mammary glands. Once the tissue is embedded in paraffin, use a rotary microtome to trim the block until the excess paraffin is removed. Next, using a vertical milling machine, drill one millimeter holes in at least two corners of the paraffin block perpendicular to the cassette mount the tissue block on the rotary microtome.

Next to set up the block face imaging system in front of the microtome, place a camera with a tele lens at a slight angle from the surface of the tissue block. The camera should be connected to a computer in front of the microtome. Then place a light source in front of the block such that the camera captures the reflection of the light from the block.

Adjust the angle of the camera until the tissue at the surface of the block is easily distinguished from the surrounding paraffin. Capture one optical block face image. Then using the microtome cut ribbons of four sections, five microns thick.

Transfer the ribbons to a cold water bath, separate the sections of the ribbon, then scoop the second and fourth sections from the cold water onto a slide. Transfer them to a warm water bath to unw wrinkle them. Choosing the second and fourth sections provides a five micron gap between sections.

Once they're unwrinkled remount, the sections on the microscope slides by scooping. As before, note that cutting, mounting and unw wrinkling the sections can cause some distortions, such as tears, folds, shrinkage, and expansion. These artifacts complicate the registration of the histology sections.

Stain the sections with h and d using an automatic stainer. Then cover slip the slides using an automatic cover slipper using a digital histology slide scanner, digitize the slides at the resolution of interest. For this protocol, the magnification is 20 x and the resolution is 0.47 microns.

This section of the video gives an overview of the image registration process, which is accomplished using custom software developed in our laboratory combined with MATLAB c plus plus. Using ITK Begin by downsampling the histology images to the resolution of the block face images. This will shrink the histology images by a factor of 0.026.

Next, for image segmentation and point selection, open a couple of block face images in MATLAB and use the data cursor tool to measure the pixel values of the registration holes. Use the average pixel value as a fixed threshold to segment the registration holes. To remove the extra segments, use circularity and the area of the segmented objects to keep the holes and remove the extra segments.

Next, for each mammary gland, select one block face image as reference. Align the centers of the registration holes of the rest of the block, face images to the centers of the registration holes of the reference image. Then using the transformations obtained for registration holes, align the block face images for the aligned block face images manually segment or extract the tissue from the background for h and d sections.

Use the OSU thresholding technique to segment images from the background and create binary masks of the histology images. Next, using the histogram of the labeled objects, identify and select the most sizable object in each mask. Extract the one pixel wide boundary points from both histology and block face masks.

Next, use a Fourier descriptors algorithm to find the initial rigid transform between the boundary points of histology and their corresponding block face images. This initial transformation includes the translation, rotation, and scale factors. Transform each histology image with the initial transform obtained from the previous step.

To refine the rigid registration, remove the high curvature edge sections from the histology contour using a rolling ball filter. Next, using uniform distribution randomly select 500 points from the remaining histology boundary points. Transform the histology boundary points with the initial transformation obtained from Fourier descriptors.

Use the block face boundary points and use iterative closest points algorithm to find the rigid transformation between the block, face boundary points and histology. Random boundary points. Then transform the histology images obtained from the previous step to create a visual image of the histology volume in the 3D visualization software.

Me vis lab, choose modules, file miscellaneous and add the composed 3D from 2D files module. Then choose image from the menu and add image property convert module. Finally, add a view 3D module from modules visualization 3D viewers.

Double click on compose 3D from 2D files to load the images and click on create 3D. Adjust the voxel size by double clicking on image property convert, and set the voxel size as 0.018 0.018 0.01. View the volume by double clicking on view 3D.

Finally, view the stack of images at five x magnification by following the steps in the accompanying document. To better characterize the deficiency of the I-G-F-B-P seven null mouse 3D reconstruction of mammary glands was performed as described herein. These figures show the reconstructed histology volumes of the wild type and null mammary glands.

Note that the mutant gland on the right is shorter and narrower than the wild type on the left, but has about the same depth with the wild type gland at 1.06 millimeters and the null gland at 1.02 millimeters deep. Also, notice that the wild type glands on the left have little stromal tissue indicated by the pink eoin staining, while the null glands on the right appear to be predominantly stromal tissue. Here the hemat toin staining, which identifies the nucleated cells, shows that the null gland maintains its density while the wild type gland appears to contain primarily glandular structures.

To further investigate this higher resolution images of the area of the lymph node were aligned here. Large structures, which would've been filled with milk are seen. In contrast, the I-G-F-B-P seven null gland has few well-developed structures.

Moreover, these structures are crowded with fibroblasts like cells. After watching this video, you should have a good understanding of how to create a 3D reconstruction of an organ through rebuilding of scan serial sections.Good. Now, while attempting this procedure, it's important to remember to consistently name the slides and scans as there will be a large number of both, and sequential organization is the key to success Following this procedure.

Other methods like immunohistochemistry for specific cell and disease markers can be performed in order to answer additional questions like, how does vasculature permeate a tumor?