تحليل حساسيت نقاط نشانه در انطباق تصاوير پزشكي

Sensitivity Analysis of Landmark Points in Medical Images Registration

Interpolation and especially thin plate spline is a popularly and widely used. algorithm for warping and registration of medical images. This interpolation function is established based on certain relationship between two sets of landmark points in source and destination planes. Sensitivity analysis of mapping function to source and for target points has not been carefully analyzed before. In this article, sensitivity analysis using analytical approach is performed on source and destination deviation. The method is applied to synthetic data, real and simulated Magnetic Resonance (MR) images. Our results demonstrate that sensitivity is highly related to landmark distribution in the source and target plane and the type of radial basis function (RBF) of the geometrical mapping. Particularly, we have shown that sensitivity is very high near landmarks that are close together and sensitivity approaches to zero near precisely extracted landmark