Deformable m-rep segmentation of object complexes

Abstract

While we have previously reported on multiscale segmentation of single-figure anatomic objects from medical images by deformable m-rep models, here we report on a method of segmentation of multiple geometrically related single-figure objects from a medical image. Inter-object geometric effects within segmentation of structures in the male pelvis from CT in the planning of radiation therapy of the prostate will be illustrated. Given a model in which each object is formed from a mesh of medial atoms, the method adds an object-complex similarity transformation stage ahead of the figural, medial atom, and boundary deformation stages of model transformation. The final three stages reward geometric interrelations among objects that occur in the model, according to the objectintrinsic coordinate systems provided by m-reps.

Computational anatomy for multi-organ analysis in medical imaging: A review

“…Based on the idea that stability provides efficiency, many works assume a predefined order of the organ analysis, usually defined heuristically, and driven by some intuitive notion of stability and/or inter-organ relations (Bloch et al, 2005(Bloch et al, , 2003Camara et al, 2004;Colliot et al, 2006;Hudelot et al, 2008;Jeong et al, 2008). Some authors solved the problem from a more quantitative approach, using image-based stability criteria, such as contrast or intensity variability, to define the order in which the organs were processed (Fletcher et al, 2002;Lu et al, 2007;Pizer et al, 2005). The anatomy of the human body has also been frequently exploited (Bloch et al, 2005;Fasquel et al, 2006;He et al, 2015;Sun et al, 2016;Udupa et al, 2013Udupa et al, , 2011Udupa and Saha, 2003;Wang and Smedby, 2014a, 2014b, using inter-organ relations, such as proximity (e.g., liver and the pancreas (Erdt et al, 2011;Hammon et al, 2013;Shimizu et al, 2010)), symmetry (e.g., left and right kidney (Camara et al, 2004)), inclusion (e.g., thoracic cavity and lungs (Camara et al, 2004;Udupa et al, 2011;Wang and Smedby, 2014b)), or intersection (e.g., hepatic vessels intersecting the liver (Fasquel et al, 2006)) to improve the computational models.…”

“…CoDeM►Han and Prince, 2003; ArtM► Anas et al, 2016;Chen et al ,2014;Chen et al, 2010;Kamojima and Miyata, 2004;Kuo et al, 2009;Martin-Fernandez et al, 2009;Miyata et al, 2003 Pelvis GlM►Li et al, 2016; CoDeM► Costa et al, 2007;Hensel et al, 2007;Ma et al, 2013;Rousson et al, 2005;Song et al, 2013; MuLeM► Chaney et al, 2013;Fletcher et al, 2002;Jeong et al, 2006;Lu et al, 2007;Merck et al, 2008;Pizer et al, 2005;SeqM► Chandra et al, 2016;Jeong et al, 2008;Ma et al, 2010;Pizer et al, 2005; MaLrM► Gao et al, 2016;Seifert et al, 2009; AtM► Akhondi-Asl et al, 2014;Weisenfeld and Warfield, 2011; Hip CoDeM► Kainmüeller et al, 2009b; MuLeM► Bukovec et al, 2011;Yokota et al, 2009;SeqM► Yokota et al, 2013; MaLrM► Glocker et al, 2012;Montillo et al, 2011; ArtM► Balestra et al, 2014;Kainmüeller et al, 2009a;Yokota et al, 2009;Knee GlM►Fripp et al, 2007; CoDeM► Pang et al, 2015;Uzumbas et al, 2013;ArtM►Constantinescu et al, 2016;...…”

“…In a different type of approach, Pizer et al (Pizer et al, 2000) used m-reps, an extension of the original medial axis descriptor (Blum, 1973) including anatomical and topological constraints. The extension to a more complex multi-organ scenario was later explored in several works (Fletcher et al, 2002;Pizer et al, 2003). However, the use of a single-organ-based parameterization remains the norm, using the inherent multi-resolution nature of m-reps to impose higher-level inter-organ constraints (see Level set functions (Samson et al, 2000) are another popular family of implicit representations of single-organ shapes, and one of the preferred methods to model multi-organ complexes via coupled deformable models (see Section 3.2).…”

A survey of prostate modeling for image analysis

“…Medial representation (m-reps) models [30] have been used to represent the 3D shape of the prostate [31]. They are composed of a set of medial atoms, which are linked together to describe an object [32]. Another nonlinear representation is the conditional shape probability distribution (CSPD) presented by Jeong et al [33].…”

Automated Segmentation of the Left Ventricle in Cardiac MRI

“…3(a),(c)). We propose to integrate this information to improve the segmentation robustness and accuracy inspired by [3,12]. Additional vertex connections are introduced that connect neighboring vertices of the existing endo-and epicardiac surface model ( Fig.…”