L'orthodontiste d'aujourd'hui est de plus en plus entraîné dans un univers tridimensionnel avec, au niveau de ses documents standards, l'apparition sans cesse plus fréquente des examens cone beam, main- tenant complétés par les empreintes numériques et les portraits 3D.

Orthodontists today work more and more in a three-dimensional (3D) universe with cone-beam examinations occurring more frequently, now supplemented by digital prints and 3D portraits.

Cephalometric analysis is the study of the dental and skeletal relationship in the head, and it is used as an assessment and planning tool for improved orthodontic treatment of a patient. Conventional cephalometric analysis identifies bony and soft-tissue landmarks in 2D cephalometric radiographs, in order to diagnose facial features and abnormalities prior to treatment, or to evaluate the progress of treatment. Recent studies in orthodontics indicate that there are persistent inaccuracies and inconsistencies in the results provided using conventional 2D cephalometric analysis. Obviously, plane geometry is inappropriate for analyzing anatomical volumes and their growth; only a 3D analysis is able to analyze the three-dimensional, anatomical maxillofacial complex, which requires computing inertia systems for individual or groups of digitally segmented teeth from an image volume of a patient's head. For the study of 3D cephalometric analysis, the current paper proposes a system for semi-automatically segmenting teeth from a cone beam computed tomography (CBCT) volume with two distinct features, including an intelligent user-input interface for automatic background seed generation, and a graphics processing unit (GPU) acceleration mechanism for three-dimensional GrowCut volume segmentation. Results show a satisfying average DICE score of 0.92, with the use of the proposed tooth segmentation system, by 15 novice users who segmented a randomly sampled tooth set. The average GrowCut processing time is around one second per tooth, excluding user interaction time.

This paper proposes a method for false-positive reduction in mammography computer aided detection (CAD) systems by detecting a linear structure (LS) in individual microcalcification (MCC) cluster candidates, which primarily involves three steps. First, it applies a modified RANSAC algorithm to a region of interest (ROI) that encloses an MCC cluster candidate to find LS. Second, a peak-to-peak ratio of two orthogonal integral-curves (named the RANSAC feature) is computed based on the results from the first step. Last, the computed RANSAC feature is, together with other MCC cancer features, used in a neural network for MCC classification, results of which are compared with the classification without the RANSAC feature. One thousand (1000) cases were used in training the classifiers, 671 cases were used in testing. The comparison shows that there is a significant improvement in terms of the reduction of linear structure associated false-positives readings (up to about 40% FP reduction).