Active volume models for 3D medical image segmentation

Abstract

In this paper, we propose a novel predictive model for object boundary, which can integrate information from any sources. The model is a dynamic "object" model whose manifestation includes a deformable surface representing shape, a volumetric interior carrying appearance statistics, and an embedded classifier that separates object from background based on current feature information. Unlike Snakes, Level Set, Graph Cut, MRF and CRF approaches, the model is "self-contained" in that it does not model the background, but rather focuses on an accurate representation of the foreground object's attributes. As we will show, however, the model is capable of reasoning about the background statistics thus can detect when is change sufficient to invoke a boundary decision. The shape of the 3D model is considered as an elastic solid, with a simplex-mesh (i.e. finite element triangulation) surface made of thousands of vertices. Deformations of the model are derived from a linear system that encodes external forces from the boundary of a Region of Interest (ROI), which is a binary mask representing the object region predicted by the current model. Efficient optimization and fast convergence of the model are achieved using the Finite Element Method (FEM). Other advantages of the model include the ease of dealing with topology changes and its ability to incorporate human interactions. Segmentation and validation results are presented for experiments on noisy 3D medical images.

Edge Detection by Adaptive Splitting II. The Three-Dimensional Case

“…However, since the model topology is created before the deformations, explicit models are not generally able to segment complex shapes with genus higher than 0. In order to overcome this limitation, several methods have been developed in past years: T-snakes [26], active volume methods [7,32], etc.…”

3D Medical Image Segmentation by Multiple-Surface Active Volume Models

“…For each independent module, the probability P r(f k |L(x)) is estimated based on the AVM's current statistics about feature f k as well as the overall feature statistics in the image [1]. Once the posterior probabilities P r(L(x)|f 1 , f 2 , ..., f n ) are estimated, we apply the Bayesian decision rule to obtain a binary map P B whose foreground represents the Region of Interest(ROI).…”

“…3D AVM [1] adopts a polyhedron mesh as the model representation which places vertices regularly on the model. More specifically, a 3D AVM is considered as an elastic solid and defined as a finite element triangulation Λ, which can be tetrahedron, octahedron or icosahedron.…”

“…In this paper, we propose Multiple-Surface Active Volume Models (MSAVM) to extract 3D objects from volumetric medical images. Being able to incorporate spatial constraints among multiple objects, MSAVM is more robust and accurate than the original Active Volume Models [1]. The main novelty in MSAVM is that it has two surfacedistance based functions to adaptively adjust the weights of contribution from the image-based region information and from spatial constraints among multiple interacting surfaces.…”

“…Active volume models (AVM) were recently proposed [1] to segment 3D objects directly from volumetric medical image dataset. Compared with active contours and ASM/AAM, the AVM is a "self-contained" generative object model that does not require off-line training.…”