Multi-modality deformable registration

Abstract

A method, system, and computer program product are provided for performing a multi-modal deformable imaging registration between two different modalities for which a functional dependency is missing or very difficult to identify. The method provides a nonlinear mapping between every pair of points in the two image modalities by calculating the nonlinear mapping for only a limited number of identifiable landmarks within the two modalities and a statistical deformation model of the structure of interest. The method comprises: calculating a mean deformation and a plurality of deformation mode Eigen vectors for a body structure from images from the same imaging mode with different deformation states for a plurality of subjects; performing a registration of a limited number of uniquely identifiable landmark points between images from two different imaging modalities for the same subject to calculate the deformation field for the landmarks from the images; determining weights for each deformation mode Eigen vector using deformation field values at the landmarks with the mean deformation and the deformation mode vectors; and summing the mean deformation field and the linear combination of the weighted deformation mode vectors to determine the deformation field between the different imaging modes for all the points within the images.

Description

technical field [0001] The present invention relates to the field of medical imaging, and more particularly to a method, system and computer program product for automatic multimodal deformable registration using statistical deformation modeling and sparse deformation data. Background technique [0002] Image registration is an important challenge in medical image processing. A major goal in medical image registration is to compute the geometric transformation that aligns the same or different images of the same object or structure. Different images can have the same modality or different modality. Common modalities for medical image registration include, but are not limited to: Magnetic Resonance Imaging (MRI), Computed Tomography (CT), and Ultrasound (US). [0003] An important subset of image registration problems involves matching images from different image modalities, sometimes referred to as multimodal image fusion. Multimodal image registration is particularly chal...

AI Technical Summary

Problems solved by technology

[0005] Lack of functional correlation between MRI and US image modalities makes it very difficult to exploit intensity-based metrics for image registration

Images