Computing deformable image registration by inverse multiquadric Demons algorithm

Radiation therapy is the mainstay of cancer treatment. However, significant weight loss and tumor shrinkage during treatment could affect the medical outcome. Advanced forms of adaptive radiotherapy use deformable image registration (DIR) accounting for such anatomical changes. Accurately defining the exact location and extent of the target with respect to surrounding organs at risk are difficult tasks. DIR model is commonly used as an alternative to measure how the tumor changes in size and location. In this paper, a reciprocal Multiquadric radial basis function approximation method is incorporated with the Demons algorithm for solving the DIR model. The proposed scheme could enhance the accuracy and efficiency of the deformable image registration. The numerical performance is compared with the classical DIR Demons algorithms. The result demonstrated that the combination of the reciprocal Multiquadric approximation and the Demons algorithm is useful in image registration.