MONICA: A Web Application for Automated Whole Optic Nerve Contour Extraction and Morphometric Analysis Validated Across Taxonomic Orders and Image Quality Levels

Abstract

Quantitative assessment of optic nerve health requires metrics beyond axon counts alone. Axon density and glial coverage fraction correlate with clinical measures of visual function, yet no existing automated tool extracts optic nerve cross-sectional boundaries to enable their calculation. We developed MONICA (Morphometrics from Optic Nerve Imaging Contour Analysis), a web application that integrates AxonDeepSeg deep learning segmentation with a novel morphology-based contour extraction algorithm to automatically derive whole nerve boundaries alongside axon and myelin masks. The contour extraction algorithm was validated against manual ground truth annotations using 15 optic nerve cross-sections spanning two taxonomic orders (mouse, rabbit), two mouse strains (BXD29, BXD51), and varying preparation quality levels (modern and archival samples). Automated contour extraction demonstrated excellent agreement with manual annotations, achieving an overall Dice similarity coefficient (a measure of segmentation overlap) of 0.987 {+/-} 0.009. Balanced precision (0.985) and recall (0.989) values indicated that the algorithm neither systematically over-segments nor under-segments nerve boundaries. MONICA requires no local software installation and runs entirely in-browser, providing batch processing for high-throughput phenotyping alongside a full suite of per-axon morphometrics. MONICA provides researchers with an accessible tool for complete nerve cross-section morphometry.