Micro-CT Imaging of a Mouse Spinal Cord

1. Mounting a Sample (Bone)

1. For examining a network of bones, like a spine, suspend the structure in an agarose gel and allow to cure in a very thin-walled plastic tube (Figure 2). The thinness of the tube is very important, greatly affecting the signal throughput and overall image quality. This in turn affects your ability to resolve features. The transmission value of the tube should be as close to 100% as possible.
2. Mount the tube on sample stage with tape or by making a custom stand, ultimately ensuring that the sample is stationary and stable when the stage rotates.

Figure 2: Mouse spine suspended in agarose gel inside thin-walled plastic tube sitting on sample stage of micro-CT system.

2. Image Acquisition

1. Turn on the X-ray source to an energy around the range of 90 keV (90 kV and 10 W).
2. After the source warms-up and settles on the energy, acquire an image through the system's software.
3. Check the transmission value by normalizing the image against an image of air. For automatic reference acquisition and application, ensure the sample can move in a given direction without crashing.
4. If image has too high of a transmission, lower the energy incrementally until the transmission value is sufficient. Be sure to increase the exposure time accordingly so that the image doesn't appear noisy and grainy. If the image has too low transmission, increase the energy incrementally until the transmission value is sufficient.
5. Begin to move the X-ray source closer to the sample, while being very careful not to crash them. Bringing the source as close as possible to the sample is a step in the direction of maximizing throughput and securing the best possible resolution.
6. As this is done, refine the field of view of the sample by shifting the sample stage with its linear actuators.
7. The software will show a parameter known as pixel size which is treated similar to the current resolution (though it is indeed different).
   1. If the number is still too large and the source is very close, the detector can begin to move away from the sample.
   2. If the number is too small, cautiously move the detector closer to the sample.
   3. Try different optical objectives and detector positions as well but beware of the complication this introduces when trying to optimize the scan parameters.
8. Check every orientation of the sample to ensure there are no crashes to find the best exposure time. This is performed because of the change of working distance, pixel size, and sample position.
9. Slowly rotate the sample by 2-degree increments while monitoring its position relative to the source and detector via the in-cabinet camera. Be sure to move the source and detector away if a collision might occur.
10. Find the longest X-ray path length that results in the lowest number of counts/transmission values and find the exposure time needed to secure about 5000 counts everywhere.

3. Tomography Submission and Reconstruction

The reconstruction process from a user perspective is not more complicated than any of the other parameter selections made in the earlier steps. However, the programming and computational expense for this process is actually quite substantial. Users must aim to best understand what is happening underneath the smooth software UI and how decisions impact the final product. Many CT systems utilize iterative algebraic reconstruction algorithms in which the 2D projections are converted into a series of linear equations describing pixel values. Some other systems utilize filtered back projection algorithms where Radon transforms convert the projections to a sinogram and are then passed through a series of line integration operations. Of course, some use other approaches and even hybrid methods. At the lowest level of involvement with these algorithms, it is known that number of projections and the total rotated displacement have an impact on the final reconstructed volume.

1. First, decide whether to scan over 180° or 360° based on the aspect ratio of the sample. If the sample has a high aspect ratio such that the X-ray path length is about 4 or more times longer at the 90°-orientation than the 0°-orientation than a 180°-scan would be a smart choice. (The argument being that the information gathered on the short X-ray path length is not that different from one side to the other. If more projections can be dedicated to the long X-ray path length orientations than the data set will benefit. The angular displacement between projections will be lower and there will more information from those tricky orientations being fed into the reconstruction algorithms.) If the aspect ratio is not that high, use 360°-scans.
2. Next, choose the number of projections and total angular displacement, which will dictate the angle between projections. The smaller this angle, the less interpolation will be conducted and less fine feature information will be truncated. (A balance must be established because more projections mean a longer scan time, which correlates to a larger window where samples can move, less time to scan other things, and a shorter source lifetime. A rule of thumb is to have at least 800 projections over 360° and to not exceed 3200 projections.
3. Submit the scan.
4. After the tomography finished (usually between 4 - 16 hours), bring 2D series of images into the system's (or some open source) reconstruction software.
5. Select the optimal center shift correction. Center shift is a parameter that aligns the projects to line up (think of a roughly shuffled deck of cards that need to be collected and aligned to sit like a neat stack). This value is usually somewhere between -10 to 10 pixels.
6. Select the optimal beam hardening correction coefficient. Beam hardening correction is an artificial removal of contrast produced by sample filtering. If a sample is thick enough or contains a range of light and heavy materials, it will have false contrast based on the attenuation of low-energy (soft) X-rays. This should be applied conservatively. An average value is somewhere between 0 - 0.5.
7. Submit reconstruction.