Generating an Image

With the knowledge of the parts of a microCT system and how X-rays interact with matter, one can begin to understand the basics of how a 3D microCT image is generated. Generating an image can be broken down into 5 general steps. These steps are outlined below with example images of a representative sample.

1. Collect a "dark field" image. A dark field image is the image collected by the detector with the X-ray beam off. It is used to calibrate for any noise from the electronics (i.e. dark current).

2. Collect a "clear field" image. A clear (or white) field image is the image collected by the detector with the X-ray beam on and the sample outside of the field of view. It is used as a calibration for any discrepancies in the intensity of the beam on the detector. In cone beam geometry, the beam is usually more intense near the center of the detector. The white field image should be collected under the same scanning conditions that the sample will be under (e.g. X-ray energy, X-ray current, detector exposure time, etc.). Essentially, it is a measurement of the initial X-ray intensity, I_o, for the Beer-Lambert Law at each pixel of the detector.

3. Collect a 2D projection image of the sample. With the proper scanning parameters chosen for a high quality scan of the sample, a projection image of the sample is collected by the detector. This image is a measurement of the attenuated X-ray intensity, I, for the Beer-Lambert Law at each pixel of the detector.

4. Rotate the sample 360 degrees while collecting more 2D projections. While the sample is rotating a full 360 degrees within the X-ray beam, a number of projections are collected at evenly spaced angles of rotation. The number of projections required for an accurate 3D image is at least equal to the number of pixel columns in the field of view. But, as a rule of thumb, the number of projections usually collected is closer to 1.5 times more than the number of pixel columns in the field of view.

5. Use a reconstruction algorithm to generate a 3D image. The most common reconstruction algorithm is Filtered Back Projection, but there a large number of reconstruction algorithms available. Basically, a reconstruction algorithm converts all of the 2D projections (calibrated by the dark and clear field images) into one 3D image. In this 3D image, the grayscale value of each 3D pixel (voxel) represents the average relative attenuation coefficient, µ, of the sample at that location in the object.