Tom Huysmans

A three-dimensional active shape model for the detection of anatomical landmarks on the back surface

Abstract In this study relations between anatomical landmarks on the dorsal surface of the human torso corresponding to underlying skeletal structures are established. By examining the statistics of the positions of the landmarks in a training set of subjects a point distribution model is derived. Rotations of the pelvis are simulated in order to show that the main mode shapes of variation are consistent with rotations of the pelvis relative to the trunk. The parameters of these mode shapes can therefore be used as independent measures of clinical parameters such as pelvic inclination, pelvic tilt, etc. The point distribution model is further applied to improve reliability and robustness for an automatic and objective detection of the anatomical landmarks on the back surface (active shape model). The results show that it is possible to replace radiographs by surface measurements in order to measure position and orientation of the pelvis, which is particularly valuable in the case of functional examinations that normally involve a large number of radiographs (e.g. to measure the position of the pelvis in a scoliosis).

A 4D-Optical Measuring System for the Dynamic Acquisition of Anatomical Structures

This paper presents a novel measuring system for the detection of moving skeletal structures. The system uses white light raster line triangulation in combination with biomechanical modeling techniques. White light raster line triangulation visualizes surfaces (e.g. the back surface) in an accurate and repeatable way, without detrimental effects, and without making contact to the human body. By making use of modeling techniques such as active contour models, active shape models and inverse kinematic models, biomechanically relevant results such as the position of the skeletal segments during motion are obtained.