Segmentation and interpolation of optic nerves in MR images

The measurement of human optic nerve is a difficult task because of its small structure. The goal of this work is to find the most suitable segmentation and interpolation models for the optic nerves in MR images, in order to improve the precision in the optic nerves area measurement. In this study, we have chosen Level Set Method (LSM) segmentation and Contrast-Guided (CG) interpolation models. We use the integrated LSM-CG model to produce distinct circular edge for the optic nerves prior the area measurement. It was found that the mean area of the optic nerve is 13.07 ± 1.29 mm
²
from original images, and 12.20 ± 1.01 mm
²
from the LSM-CG processed images, which are within the range of the reported literature values. These measurement were obtained from Proton Density (PD) images from five healthy volunteers on both 1.5T and 3.0T MRI scanners. The integrated LSM-CG model is suitable for processing optic nerve images to achieve higher resolution edges and allow more accurate area measurement.