Constructing average models of quasi-spherical objects: application to corneal topographies. Issue 4 (4th July 2017)

• Record Type: Journal Article
• Authors: Polette, Arnaud; Auvinet, Edouard; Mari, Jean-Luc; Brunette, Isabelle; Meunier, Jean
• Is Part Of: Computer methods in biomechanics and biomedical engineering. Volume 5:Issue 4(2017)
• Page Start: 241
• Page End: 250
• Publication Date: 2017-07-04
• Subjects: Average models -- surface -- registration -- atlas -- cornea; Imaging systems in biology -- Periodicals; Imaging systems in medicine -- Periodicals; Biomechanics -- Data processing -- Periodicals; Biomedical engineering -- Periodicals; 616.0757
• DOI: 10.1080/21681163.2015.1077164
• Languages: English
• ISSNs: 2168-1163

Abstract

Abstract : In medical imaging, it is now common to create 3D models of organs by 'averaging' several specimens obtained from different subjects. This requires a registration step to align the organs before averaging their shapes. In this paper, we present the difficult case of a quasi-spherical organ: the cornea. To cope with the lack of anatomical anchor points, we use a registration algorithm based on the minimisation of a global factor: the volume between the two surfaces to be registered. The cornea is a thin tissue layered by two (anterior and posterior) surfaces. Therefore, we actually introduce a third virtual surface to drive the two others. After registration using an iterative optimisation algorithm, anterior and posterior average surfaces are computed. Our study demonstrates that this matching step is crucial to correctly build and compare surfaces. Several clinical applications of this methodology are also presented to illustrate its efficiency.