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

Abstract

Abstract. 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. These two functions help MSAVM not only overcome local minima but also avoid leakage. Because of the implicit representation of AVM, the spatial information can be calculated based on the model's signed distance transform map with very low extra computational cost. The MSAVM thus has the efficiency of the original 3D AVM but produces more accurate results. 3D segmentation results, validation and comparison are presented for experiments on volumetric medical images.

Multi-Layer Deformable Models for medical image segmentation

“…The proposed geometry makes self-intersections less probable compared to other volumetric models. In contrast to [9], [8], shape priors can be easily integrated since every surface point has corresponding points in the inner layers. We developed an corresponding deformation logic that combines such local shape and intensity priors with the volumetric information gained from samples inside of the model (cf.…”

“…Barreira et al [8] proposed a volumetric model whose interior consists of a 3D grid that can be torn apart in order to segment multiple structures in one step. Recently, active volume models (AVM) [9] have been proposed, which use a finite element triangulation to guide a surface model. However, in these approaches the model's interior structure is not symmetric to the outer tessellation.…”

Synthesis of intensity gradient and texture information for efficient three-dimensional segmentation of medical volumes

“…17 was enhanced by Shen et al in Ref. 18, where a multiple-surface AVM was proposed to simultaneously segment coupled medical objects. Olowoyeye et al 19 pioneered a medical (CT and MRI) volume segmentation methodology using its textural properties derived via a 3-D Gabor filter bank whose number depended on the size of the volume.…”

Active Volume Models for Medical Image Segmentation

“…We further propose the Multiple-Surface AVMs (MSAVM) [32] to segment multiple coupled objects simultaneously. Instead of setting up fixed inter-surface distance constraints during initialization, MSAVM dynamically updates the distance constraints between the interacting surfaces based on current model surfaces' spatial interrelations.…”

“…Therefore, a single surface based deformable model may stop at local minima or leak out to incorrectly converge at a nearby object's boundary. Often such mistakes can be avoided by considering spatial constraints between multiple objects -for instance, by integrating the spatial constraints in a multiplesurface based deformable model framework [7], [32], [42] and deforming all interacting surfaces simultaneously to extract the object boundaries with better accuracy.…”