Heart Reconstruction Pipeline

2. Pipeline


Data Flow for Modeling Patient-Specific Human Heart.

Our IMAGING2MODELING framework consists of several image processing, and geometric processing algorithms. The overall framework is built with two main processing units – Image Processing and Geometry Processing. The image processing unit has the modules for contrast enhancement, denoising, image segmentation and contouring/reconstruction. The geometry processing unit consists of curation and geometry segmentation,  skeleton extraction, muscle wall construction, denoising and decimation. The mesh generation unit is also part of geometry processing and it has its own modules for linear and higher order surface and volumetric mesh generation. While the rest of the algorithms are known and can be found in the corresponding papers cited therewith, here we focus on the steps of geometric curation, geometric segmentation, muscle wall construction, and placement of coronary arteries and veins to build a complete meshed model of a patient’s heart.


Reconstruction pipeline. (a) A sinlge slice of high resolution volumetric CT. (b) Reconstructed volume. (c) Raw initial geometric model. (d) Refined model after curation and segmentation