ATAA biomechanics analysis workflow - lianggecm/ATAASegmentation Wiki

Welcome to the ATAASegmentation wiki!

We are working on biomechanics of ascending thoracic aortic aneurysms. The typical workflow of the project starts from the contrast-enhanced computer tomography angiography (CTA) images of a patient's chest. The 3D volume images are fed into a pre-trained neural network to create a pixel-wise segmentation of the ascending aorta.

The segmented images from the neural network are then used to extract the 3D surface of the aortic aneurysm using the marching cube algorithm.

The surface geometry created from this step is highly irregular and non-smooth and needs to be smoothed before it can be used for biomechanics analysis (otherwise would lead to non-physical solution). We use a radial-basis function based smoothing algorithm to create a smoothed representation of the ascending aorta:

Once the smoothed geometry is created, we use an optimization algorithm to automatically create a 3D finite element mesh as shown here: