Deep-learnt tissue deformation for medical imaging

Abstract

A deep machine-learning approach is used for medical image fusion (24) by a medical imaging system (48). This one approach may be used for different applications. For a given application, the same deep learning is used but with different application-specific training data. The resulting deep-learnt classifier provides a reduced feature vector in response to input of intensities of one image and displacement vectors for patches of the one image relative to another image. The output feature vector is used to determine (16) the deformation for medical image fusion (24).

Description

Background technique [0001] This embodiment relates to medical imaging. Blending medical images together to assist in the diagnosis and / or treatment of patients. For medical image fusion, including unimodal and multimodal, deformable registration is often used to compensate for changes in body position, organ deformation, and / or cardiac and respiratory motion between the images to be fused and peristaltic motion. Deformable registration first matches the corresponding image structure by a similarity measure, and then bends one of the images based on a physical or mathematical model to ensure that the deformation is consistent with the actual human body changes. Deformable registration can use elastic or fluid modeling, optical flow, biomechanical modeling, or diffeomorphism. [0002] Deformable models attempt to make the deformation smooth across space so that adjacent parts do not have unrealistic movement relative to each other. Biomechanical properties can be applied to...

AI Technical Summary

Problems solved by technology

Using deep machine learning methods

This approach can be used for different applications

Images