Incompressible Biventricular Model Construction and Heart Segmentation of 4D Tagged MRI.

Most automated methods for cardiac segmentation are not directly applicable to tagged MRI (tMRI) because they do not handle all of the analysis challenges: tags obscure heart boundaries, low contrast, image artifacts, and radial image planes. Other methods do not process all acquired tMRI data or do not ensure tissue incompressibility. In this chapter, we present a cardiac segmentation method for tMRI which requires no user input, suppresses image artifacts, extracts heart features using 3D grayscale morphology, and constructs a biventricular model from the data that ensures the near incompressibility of heart tissue. We project landmarks of 3D features along curves in the solution to a PDE, and embed biomechanical constraints using the finite element method. Testing on normal and diseased subjects yields an RMS segmentation accuracy of ~2 mm, comparing favorably with manual segmentation, interexpert variability and segmentation methods for nontagged cine MRI.