Glaucoma, a leading cause of blindness, results in progressive vision loss if untreated. Optical coherence tomography (OCT) enables the measurement of retinal nerve fiber layers and the optic nerve head (ONH). Považay et al. introduced the Pigment epithelium central limit-Inner limit of the retina Minimal Distance averaged over 2π radians (PIMD-2π) to quantify the minimal cross-section of the nerve fiber l ayer i n the ONH. The present research enhances automated PIMD estimation in OCT images, employing the nnU-Net model for semantic segmentation. Using a dataset of 78 OCT images from Uppsala University, experiments were conducted in cylindrical (2D U-Net and nnU-Net) and Cartesian domains (nnU-Net). Results show that the nnU-Net frameworks significantly improve OPCL coordinate accuracy (mean Euclidean distance in pixel value: 1.665 for cylindrical and 2.4495 for Cartesian) compared to 2D U-Net (10.6827). Notably, the nnU-Net Cartesian architecture removes manual bias from ONH center selection during cylindrical transformations. PIMD calculations effectively distinguished glaucoma patients from healthy subjects, with nnU-Net methods demonstrating superior stability. This study underscores the potential of automated PIMD estimation in advancing glaucoma .