A least squares approach for the reconstruction of nerve fiber orientations from tiltable specimen experiments in 3D-PLI

3D-Polarized Light Imaging has become a unique technique to study the fiber architecture of unstained brain sections at the meso- and microscale. It exploits the intrinsic birefringence of nerve fibers which is measured with a customized Polarimeter in which the brain section is placed on a tiltable specimen stage. So far, a computationally fast analytical method based on the discrete Fourier transformation to analyze the data acquired with the tiltable specimen stage has been used. In this study, we propose a new algorithm based on a fitting approach which provides an improved stability against measurement noise resulting in a more realistic orientation interpretation, in particular for low signals. For the first time, it is demonstrated how fiber courses at the boundary of white and grey matter can robustly be reconstructed with 3D-PLI. This significantly improves the reliability of mapping the cortex based on 3D-PLI data.