Automated image assessment of posterior capsule opacification using Hölder exponents

Posterior Capsule Opacification (PCO) remains to be the most common complication of cataract surgery after intraocular lens implantation. Though several strategies have been suggested for the prevention of PCO, a standard PCO quantification system is required to reliably assess the effectiveness of these strategies. This paper proposes a method based on computation of Hölder exponents to quantify the amount of PCO in the digital image. PCO areas are effectively detected and classified according to their severity using histogram-based thresholding on Hölder exponent image. This method is implemented in Matlab and verified on real PCO images. The results show a high correlation of 83% between the computed PCO scores and the clinical grades, as well as demonstrate the robustness of the proposed system to monotonic illumination variations.