Optimal Selection of Operating Point for Hessian-based Polyp Detection Method in CT Colonography

摘要

Hessian matrix is the square matrix of second partial derivatives of a scalar-valued function and is well known for object recognition in computer vision and medical shape analysis. We appied this Hessian matrix to polyp detection in CTC. This Hessian-based methods requires two optional parameters, Gaussian blurring factor and shape threshold. To optimize the parameters and validate this method, we have produced anthropomorphic pig phantoms. We performed FROC analysis on these phantoms to find optimal shape threshold in regarding high sensitivity and relatively low false positive rate. We found 110 as optimal shape threshold and in this setting we got 82.5% sensitivity, 8.64 as FP rate.