Bayesian-Optimized Superpixel-GrabCut for Traceable Optic Disc Segmentation
Computer Vision and Pattern Recognition
Summary
The authors created a new way to find the optic disc in eye images that helps doctors diagnose eye diseases. Unlike many existing methods that work like a mystery box, their approach shows clear, step-by-step math for how decisions are made. They combined several techniques, like breaking the image into small pieces and fitting shapes, and tweaked settings automatically to get the best results. Their method performed just as well as top current methods while being easier to understand and check.
Authors
Shraddha Changune, Vivek Noel Soren, Gautam Das, Tapan Kumar Gandhi
Abstract
Optic disc (OD) segmentation is essential for diagnosing ophthalmic pathologies from retinal fundus images. However, prevailing deep learning approaches operate as opaque black boxes, lacking the inference-stage mathematical traceability--a critical requirement for algorithmic auditing and failure analysis in clinical workflows. This paper presents a fully algorithmically traceable and trainable segmentation pipeline that jointly combines superpixel decomposition, hybrid brightness-proximity superpixel scoring, morphological regularization, iterative GrabCut refinement, and elliptical shape fitting. The hyperparameter optimization is formulated as an objective function and solved via Bayesian optimization to eliminate manual parameter tuning. A quantitative evaluation on the Drishti-GS dataset demonstrates that our method achieves a Dice coefficient of 0.9536, matching state-of-the-art performance. By maintaining explicit mathematical transparency across all processing stages, our framework offers a deterministic, traceable alternative to black-box architectures for medical review and debugging.