Multiobjective optimization framework for landmark measurement error correction in three-dimensional cephalometric tomography

• Published in: Dento-maxillo-facial radiology Vol. 42; no. 7; p. 20130035
• Main Authors: DeCesare, A, Secanell, M, Lagravère, MO, Carey, J
• Format: Journal Article
• Language: English
• Published: England The British Institute of Radiology 01.07.2013
• Subjects: Algorithms; Anatomic Landmarks - anatomy & histology; Anatomic Landmarks - diagnostic imaging; Cephalometry - methods; Cone-Beam Computed Tomography - methods; Ear Canal - anatomy & histology; Ear Canal - diagnostic imaging; Facial Bones - anatomy & histology; Facial Bones - diagnostic imaging; Foramen Magnum - anatomy & histology; Foramen Magnum - diagnostic imaging; Humans; Image Processing, Computer-Assisted - methods; Imaging, Three-Dimensional - methods; Mandible - anatomy & histology; Mandible - diagnostic imaging; Orbit - anatomy & histology; Orbit - diagnostic imaging; Palatal Expansion Technique; Reproducibility of Results; Retrospective Studies; Skull - anatomy & histology; Skull - diagnostic imaging; Skull Base - anatomy & histology; Skull Base - diagnostic imaging; Sphenoid Bone - anatomy & histology; Sphenoid Bone - diagnostic imaging; orthodontics; landmark optimization procedure; cone beam computed tomography; maxillary expansion treatment
• ISSN: 0250-832X; 1476-542X; 1476-542X
• DOI: 10.1259/dmfr.20130035

The purpose of this study is to minimize errors that occur when using a four vs six landmark superimpositioning method in the cranial base to define the co-ordinate system. Cone beam CT volumetric data from ten patients were used for this study. Co-ordinate system transformations were performed. A co-ordinate system was constructed using two planes defined by four anatomical landmarks located by an orthodontist. A second co-ordinate system was constructed using four anatomical landmarks that are corrected using a numerical optimization algorithm for any landmark location operator error using information from six landmarks. The optimization algorithm minimizes the relative distance and angle between the known fixed points in the two images to find the correction. Measurement errors and co-ordinates in all axes were obtained for each co-ordinate system. Significant improvement is observed after using the landmark correction algorithm to position the final co-ordinate system. The errors found in a previous study are significantly reduced. Errors found were between 1 mm and 2 mm. When analysing real patient data, it was found that the 6-point correction algorithm reduced errors between images and increased intrapoint reliability. A novel method of optimizing the overlay of three-dimensional images using a 6-point correction algorithm was introduced and examined. This method demonstrated greater reliability and reproducibility than the previous 4-point correction algorithm.