Calibration of 3D-PLI measurements

Background and Motivation

The Three-dimensional Polarized Light Imaging (3D-PLI) technology is used to capture high resolution images of thinly sliced sections of post-mortem brains, using a polarizing microscope. These images, if available in adequate quality, can then be stacked to reconstruct the brain in three dimensions. Such a 3D reconstruction enables tracking of individual nerve fibers through the entire brain.

However, during the imaging process, the quality of each image is affected by different types of effects, such as image noise, e.g., vignetting, irregular illumination, etc. These effects have a direct impact on the reconstruction of the nerve fibers. Therefore a precise calibration of the raw data is important for further processing.

Our approach

By using flat-field images, i.e., images of the background without any brain tissue, a calibration matrix for 3D-PLI raw data is calculated and applied (see Figure 1). We use statistical methods to reduce the number of required flat-field images to calculate the calibration matrix while maintaining a sufficient level of stability and quality. The results of these improvements are shown in Figure 2.

Figure 1: The image on the top left depicts the uneven illumination typically seen in a flat-field image. Calibration of this image with the calibration matrix, i.e., the image on the top right, results in the image at the bottom, which is evenly illuminated.
Figure 1: The image on the top left depicts the uneven illumination typically seen in a flat-field image. Calibration of this image with the calibration matrix, i.e., the image on the top right, results in the image at the bottom, which is evenly illuminated.

Furthermore, our tool can continuously update the calibration matrix using flat-field images that are captured before every normal measurement.

Our implementation of the calibration algorithm is parallelized and optimized to minimize the required time-till-solution. The application can utilize the supercomputing resources to significantly increase the speed of the calibration process.

Calibration of 3D-PLI measurements
Figure 2: Histogram showing the distribution of intensity values across all pixels in an image. The narrower the distribution, the more evenly illuminated is the image. As depicted in the figure, our optimizations to the calibration process result in the narrowest distribution.

Our collaboration partners

This project is being conducted in collaboration with the Fiber Architecture group of the INM-1.

Simlab Contact

Kim Sontheimer

Last Modified: 06.05.2022