Abstract— In this paper a method based on mesh surfaces approximations for the 3D analysis of anatomical structures in Radiotherapy (RT) is presented. Parotid glands meshes constructed from Megavoltage CT (MVCT) images were studied in terms of volume, distance between center of mass (distCOM) of the right and left parotids, dice similarity coefficient (DICE), maximum distance between meshes (DMax) and the average symmetric distance (ASD). A comparison with the standard binary images approach was performed. While absence of significant differences in terms of volume, DistCOM and DICE indices suggests that both approaches are comparable, the fact that the ASD showed significant difference (p=0.002) and the DMax was almost significant (p=0.053) suggests that the mesh approach should be adopted to provide accurate comparison between 3D anatomical structures of interest in RT. I. INTRODUCTION HE analysis of anatomical structures of interest is a relevant task in the medical field. For example in Radiotherapy (RT), patients undergoing a head-and-neck cancer (HNC) treatment are known to experience significant decrease in the volume of the parotid glands and their migration toward the midline of the patient with a distance change of a few millimeters. Due to these anatomical modifications the parotids can receive a total dose significantly higher than the planned one. In this context the importance of the analysis of the anatomical modifications occurring during RT treatments is to both identify potential predictors of toxicity and design optimized adaptive treatment plans [1]. *A. R. Mejia-Rodriguez is with Institute of Molecular Bioimaging and Physiology (IBFM)-CNR, Milan, Italy and Bioengineering Department, Politecnico di Milano, Milan, Italy. Author is supported by CONACyT studentship (CVU/Becario): 217232/213579 for PhD studies (e-mail: armero83@gmail.com) E. Scalco is with Institute of Molecular Bioimaging and Physiology (IBFM)-CNR, Milan, Italy D. Tresoldi is with Institute of Molecular Bioimaging and Physiology (IBFM)-CNR, Milan, Italy and Bioengineering Department, Politecnico di Milano, Milan, Italy. A.M. Bianchi is with Bioengineering Department Politecnico di Milano, Milan, Italy. E. R. Roman-Arce is with Fac. de Ciencias, UASLP, San Luis Potosi, Mexico. M.O. Mendez-Garcia is with Fac. de Ciencias, UASLP, San Luis Potosi, Mexico. G. Rizzo is with Institute of Molecular Bioimaging and Physiology (IBFM)-CNR, Milan, Italy. To have a better perspective of the distortions that occur in structures of interest, different techniques to represent structures in 3D have been implemented [2, 3]. The mesh data-structure is one of the most used techniques for rendering 3D objects and it is defined as a collection of vertices (points positioned in a virtual space), edges (a connection between two vertices) and faces (a closed set of edges) that defines the shape of a polyhedral object. The faces could form polygons of any type (quadrilaterals, concaves or convexes complex polygons) but triangular polygons are commonly used since this simplifies rendering [3]. Although meshing is widely used for rendering 3D geometrics, in RT, analysis of anatomical structures of interest is usually carried out by extracting quantitative measurements from a set of 2D binary images whose contours define a 3D surface [4, 5]. In this way a coarser representation of the structure surfaces is obtained [2]. In this paper, we propose a mesh-based approach to improve the analysis of 3D anatomical structures in RT. This approach is compared with the standard binary image approach in order to evaluate if the introduction of this more accurate 3D structure representation results in differences in structure analysis. II. METHODS A. Analysis of structures of interest by meshes Analysis of the 3D anatomical structures by meshes proposed in this paper consists in the calculation of nine quantitative standard indices. The indices could be divided into two categories, 1) indices that give spatial and geometrical information of an individual mesh: coordinates of the baricenter, surface area and volume; and 2) indices that make a comparison between two meshes: Euclidean distance between baricenters, surface area difference, volume difference, dice similarity coefficient (DICE) [1], maximum distance between meshes (DMax) and average symmetric distance (ASD) [1]. DICE index is a metric of the overlap between 2 surfaces ranging from 0 (no spatial overlap) to 1 (complete overlap). DMax index calculates the maximum distance between two surfaces and gives an idea of the worst local distance mismatch. ASD index is defined as the average Euclidean distance between two surfaces, which is 0 for a perfect match. A Mesh-Based Approach for the 3D Analysis of Anatomical Structures of Interest in Radiotherapy Aldo R. Mejia-Rodriguez , Elisa Scalco, Daniele Tresoldi, Anna M. Bianchi, Edgar R. Arce-Santana, Martin O. Mendez, Giovanna Rizzo T 34th Annual International Conference of the IEEE EMBS San Diego, California USA, 28 August - 1 September, 2012 6555 978-1-4577-1787-1/12/$26.00 ©2012 IEEE