利用3D打印技术来合并MRI与组织学：一个大脑切片协议

The overall goal of this protocol is to accurately align magnetic resonance imaging image volumes with histology sections via that creation of customized 3D printed brain holders and slicer boxes. While this video focuses on brain tissue extracted from marmoset, this protocol can be applied to any brain whether it's coming from a rodent, a non-human primate or human. Prepare a workstation as described in the protocol.

Fill the tube with Fomblin and gauze to the 20 milliliter mark. Compress the gauze to remove air bubbles along the way. Gently the dry the brain of formalin.

Then insert the brain with the front towards the bottom of the tube. Carefully secure the brain in the tube using gauze around the side of the brain to fix its position. Fill the rest of the tube with gauze and Fomblin.

Carefully remove air bubbles along the way. Then secure the cap and seal the tube with paraffin. Mark the cap and line with the inter hemisphere fissure.

Then wrap the tube in a paper towel and insert it into the coil with the mark at the top center position. Insert the coil into the MRI scanner. After the acquisition and post processing, we have a T2 weighted image we can use to make a digital model of the brain.

Open the MRI in the coronal orientation. Then use the transformation tools to resample the image to 1 millimeter isotropic voxels. Next, apply a nonlinear noise reduction filter.

Select displays look up table. Then click on the dual Threshold button. Drag the slider on the graph to cover the whole brain.

Create a binary brain mask using the Segmentation algorithm Threshold using min/max. Enter the value located in the bottom left corner of the intensity graph, just below the scale into the Lower limit box. Then select Binary and uncheck Inverse threshold.

Select the Morphological algorithm, Fill holes. Check Process in 2.5D. To fill in the gap between the hind brain and the cortex, select line VOI.

Draw in a connection between the hind brain and the cortex on both sides of the brain at the most lateral point. Continue drawing VOI connections through the brain. Convert the VOIs to a Binary mask.

Use the Image calculator to combine the VOI and brain Binary masks by selecting the OR operation. Select Promote destination image type. Now select the Morphological algorithm, Fill holes.

And check Process in 2.5D. Select the algorithm Extract surface Select Mask image, enter a name and then select the file type ply. The gaps in the cutting box corresponding to the slice plane location can be placed starting from the front of the brain, the back of the brain or relative to an area of interest such as the lesions in the white matter seen here.

Using box VOI, draw a box around the brain. Copy and paste the box onto the number of MRI slices corresponding to the intended plate gaps thickness. Then skip ahead by the number of MRI slices per section and piece the box as done before to create another blade gap.

Repeat this process through the brain. Invert the box VOIs to a Binary mask. Then extract the surface using Extract surface as done previously for the brain.

Select the blade gap's Binary mask. Use Image math to multiply it by 10, 000. Then use Image calculator to add the multiplied Binary mask to the resampled isotropic MRI.

Now by selecting the triplanar view, an MRI blade map showing the location the brain will be sliced can be seen in three orthogonal views. Netfabb Professional, select Add part. Then select the Blade Gap model and Brain Model plys.

Select the brain model and click on repair mode. Use the shell selection button to select the brain. Then toggle the selection to the other shells and delete them.

Then apply the repair and remove the old part. With the brain selected, select Move. Click on to origin and record the parameters in the X, Y, and Z boxes.

These are needed to maintain blade gap positioning. Click on translate. And close the window.

Select the Blade Gap model and translate it using the parameters recorded from the previous step. Select the Brain model and click on repair mode. Use the shell selection button to select the brain.

Then select Smooth triangles. Enter four, for the number of Iterations. And check Prevent volume shrinking.

Select Reduce Triangles and enter 200000 in the Target Triangle Count box. Then select Automatic repair and perform a default repair. Apply the repair.

And then export the smoothed brain model as an STL. Import smoothed brain model into Meshmixer. Use the sculpting tools to smooth the areas drawing with the line VOIs.

And fix any other minor defects with the model. Import the smoothed, edited brain model into Netfabb Professional. Then add the part, brain slicer parts marmoset.

Select Shells to parts. Then rename each part as described in the protocol. Using the selection settings described in the protocol, click and drag the parts to position the brain in the center of the box.

Click and drag the parts to adjust the depth of the brain in the box. Select the smoothed, edited brain model, in the box cut out model and select boolean operation. Click on the brain model to turn it red.

And select boolean subtraction. Apply the calculations. Check the box for overhangs that will prevent the brain from being properly placed.

Refer to the supplementary section of the protocol for a method to remove them. Select the blade gap model and click move. Record at Z position.

This will be the position of the most posterior blade gap. Then close the window. Select the blade part from the brain slicer parts and click on Move.

Enter the Z value from the previous step. And enter the current position X and Y values. Select absolute Translation and translate.

Click on Duplicate. If the section thickness is consistent, as shown here, enter the total blade number and the total count and Z count, and enter the section thickness in the Z gap box. Then select duplicate.

Select the microtone blade part from the brain slicer parts. Repeat the same moving and duplicating steps as done with the blade part. Select all the Microtome Blades and the main blade holder and click on boolean operation.

Select all the Microtome Blades to turn them red and click on boolean subtraction. Then apply the calculations. Enter repair mode and select Automatic repair.

Perform an Extended Repair and apply the repair. Export the blade holder as an STL. Select all of the blades, the main and sub boxes, and the smoothed, edited Brain model, and click on boolean operation.

Use the Parts selection window to turn all the Parts, except for the main box, red. Then select boolean subtraction. Apply the calculations.

Enter repair mode and check the box for sharp edges that could injure the brain. These can be repaired in Netfabb repair mode. Or using the sculpting tools in Meshmixer.

Then select Automatic repair. Perform an Extended Repair. Then apply the repair.

Export the brain box as an STL. In Cura, select Load and import the Brain Box. Click on the Rotate button.

Then click and drag to lay the box flat. Edit the print settings as shown in the protocol. And then Save the toolpath.

Follow the same process for the blade holder and also duplicate the object. Then Save the toolpath. After applying a thin layer of glue to the print bed, print the brain box and blade holders on the 3D printer using the settings described in the protocol.

After the prints complete, the brain can be prepared for cutting. Set up a workstation as described in the protocol. Carefully position the brain in the box and make sure it is firmly in place.

Position the Microtome Blades in the blade holder into the corresponding gaps in the box. Wearing protective gloves, push down on the blade holders firmly. Applying slow, balanced pressure to cut the brain.

Remove the slabs one at a time starting from the front of the brain. Although this approach is straight forward, the method requires many steps, as well as the use of several types of software and investigators should familiarize themself with user interfaces before starting the procedure. This workflow summarizes the protocol for making the Marmoset brain slicer box.

The brain is fixed with Formalin and a T2 weighted MRI is acquired at 150 micron. Images are processed and thresholded to create a Binary mask. The surface is then rendered in 3D modeling software.

A boolean subtraction between a slicer template and the brain model create the digital model of the brain slicer. The brain slicer box is then printed on a 3D printer. The brain is then placed firmly in the slicer box for cutting.

Once the brain is sliced, a visual comparison between the In Vivo and post-mortem MRI images and pictures of the superficial surfaces of the slabs reveals a good orientation match across multiple slabs. A more thorough comparison between the high resolution post-mortem MRI and the LFB CV stained histology sections from multiple brain slabs demonstrate an accurate and consistent match across all the structures of the marmoset brain. In this animal model of Multiple Sclerosis, the animals develop white matter lesions spread throughout the brain.

Using this technique, small lesions detected on In Vivo MRI can be tracked on both post-mortem MRI and histology sections. The histology and post-mortem MRI can also reveal lesions otherwise missed on In Vivo MRI. Here, the In Vivo MRI did not show convincing evidence about normal, hyperintense signal to suggest lesions in either optic tract.

Nevertheless, in the post-mortem MRI, signal hyperintensities in the optic tract were observed. And these areas corresponded directly with the myelinated lesions seen in the LFB CV histology section. So in conclusion, the methodology introduced here allows investigators to accurately assess the pathology underlying MRI findings.

We think it's a promising approach for identifying novel biomarkers for diseases involving the brain and for specific pathological processes such as inflammation and remyelination.