Alert button

Deep Active Latent Surfaces for Medical Geometries

Jun 21, 2022
Figure 1 for Deep Active Latent Surfaces for Medical Geometries
Figure 2 for Deep Active Latent Surfaces for Medical Geometries
Figure 3 for Deep Active Latent Surfaces for Medical Geometries
Figure 4 for Deep Active Latent Surfaces for Medical Geometries

Share this with someone who'll enjoy it:

Shape priors have long been known to be effective when reconstructing 3D shapes from noisy or incomplete data. When using a deep-learning based shape representation, this often involves learning a latent representation, which can be either in the form of a single global vector or of multiple local ones. The latter allows more flexibility but is prone to overfitting. In this paper, we advocate a hybrid approach representing shapes in terms of 3D meshes with a separate latent vector at each vertex. During training the latent vectors are constrained to have the same value, which avoids overfitting. For inference, the latent vectors are updated independently while imposing spatial regularization constraints. We show that this gives us both flexibility and generalization capabilities, which we demonstrate on several medical image processing tasks.

* 14 pages, 9 figures, submitted for review  

Share this with someone who'll enjoy it: