Optimised Misalignment Correction from Cine MR Slices Using Statistical Shape Model

Cardiac magnetic resonance (CMR) imaging is a valuable imaging technique for the diagnosis and characterisation of cardiovascular diseases. In clinical practice, it is commonly acquired as a collection of separated and independent 2D image planes, limiting its accuracy in 3D analysis. One of the major issues for 3D reconstruction of human heart surfaces from CMR slices is the misalignment between heart slices, often arising from breathing or subject motion. In this regard, the objective of this work is to develop a method for optimal correction of slice misalignments using a statistical shape model (SSM), for accurate 3D modelling of the heart. After extracting the heart contours from 2D cine slices, we perform initial misalignment corrections using the image intensities and the heart contours. Next, our proposed misalignment correction is performed by first optimally fitting an SSM to the sparse heart contours in 3D space and then optimally aligning the heart slices on the SSM, accounting for both in-plane and out-of-plane misalignments. The performance of the proposed approach is evaluated on a cohort of 20 subjects selected from the UK Biobank study, demonstrating an average reduction of misalignment artifacts from 1.14 ± 0.23 mm to 0.72 ± 0.11 mm, in terms of distance from the final reconstructed 3D mesh.