An automated segmentation and biometry method for mouse axial OCT images based on Longitudinal Intensity Profile Analysis (LIPA) in mouse.

This study aimed to develop an automated segmentation and measurement algorithm for axial mouse optical coherence tomography (OCT) images to enable precise identification of ocular structures and quantification of their geometric parameters. The Longitudinal Intensity Profile Analysis (LIPA) algorithm was developed based on anatomical prior knowledge and a hierarchical cascaded search. Utilizing ocular images obtained from male C57BL/6 mice using two independent swept-source OCT systems, the algorithm localized the region of interest, constructed a longitudinal intensity profile to enhance tissue boundaries, and sequentially located five key interfaces within dynamically constrained search windows to automatically calculate component-specific axial lengths incorporating respective refractive indices. Validation against manual measurements on both OCT datasets using intraclass correlation coefficient (ICC), Pearson correlation, and Bland-Altman analysis demonstrated that the algorithm achieved very strong correlations and high agreement for total axial length and its components (all Pearson's r > 0.900, ICC > 0.900 on the primary dataset), maintaining high accuracy for key parameters like lens thickness on the secondary device (all Pearson's r > 0.950, ICC > 0.950). The results confirm strong agreement between the algorithm and manual measurements across datasets. In conclusion, the LIPA algorithm provides an efficient, reproducible, and generalizable tool for automated ocular biometry in mice, suitable for large-scale, standardized research in ocular development and myopia.