Anterior Segment OCT Image Segmentation Correction

The patent divides the image processing into multiple stages: first identifying a candidate region using initial segmentation, then performing refined segmentation only within that candidate region. This multi-level segmentation approach allows automatic processing while improving accuracy by limiting the refined segmentation to relevant areas, reducing the impact of artifacts in other regions.

The patent applies different processing quality to different regions of the image. The candidate region identification provides a coarse segmentation for most areas, while refined segmentation with higher precision is applied only to the specific candidate region where corneal surfaces are expected. This local quality approach maintains automation while improving measurement precision where it matters most.

The patent segments the processing workflow into distinct functional modules: a candidate region identification module that performs initial filtering, and a refined segmentation module that applies precise edge detection and curve fitting only to the identified candidate regions. This modular segmentation of processing steps improves accuracy while keeping each module relatively simple and independent.

The patent performs preliminary candidate region identification before executing the more complex refined segmentation algorithms. This preliminary action filters out large portions of the image that do not contain corneal surfaces, allowing the computationally intensive refined segmentation to be applied only where necessary, thus improving accuracy without proportionally increasing overall system complexity.