• Michael Schaich
  • Florian Dammann
  • Esther Grabert
  • Jörg Subke
  • Thomas Horstmann
  • Claus D. Claussen
Keywords: segmentation, biomechanics, magnetic resonance tomography


INTRODUCTION: This work is part of a project of the Department for Sports Medicine to calculate the internal stresses arising when jumping from a squat position. The goal of the project is to facilitate individual calculations by establishing a biomechanical model whose parameters are the major anatomical-geometrical and physiological quantities, gained by electromyogram (EMG) and radiological measurement. Procedures for acquiring the latter data are described here. METHODS: As the study did not involve pathologies, ionizing radiation was ruled out, and magnetic resonance imaging (MRI) was used. The biomechanical model required geometrical parameters from joint positions beyond those occurring during the squat-vault, so the Siemens Magnetom Open device was chosen. It has the disadvantage of relatively low magnetic field strength (0.2 T), but allows for almost unlimited movement in the table plane. Different measurement parameters were evaluated. As the length of the field of view was about 25 cm, the different joints had to be scanned separately. A positioning table was used to serve three purposes: 1. Positioning with defined joint angles, so the morphology could be related to the EMG measurements. 2. Exertion of force, to measure the geometry of muscles and tendons under strain. 3. Placement of markers with high MRI contrast, to relate the relative position of the scans of the different joints. [delete line space]. The evaluation of the images was done using the ‘Tübinger Medstation’ software developed by the Department of Computer Science at the University of Tübingen. RESULTS: Although the use of T2 weighted sequences resulted in better soft tissue contrast, the T1 weighted spin echo sequence was preferred because of shorter acquisition time, which was an important factor because measurements had to be made under strain. Bones and tendons, with their low hydrogen content, produce weak signals in MRI and thus contrast with the adjacent soft tissue. Even shorter acquisition times by use of a gradient sequence were ruled out because of their low signal/noise ratio, which rendered the fascies undetectable. Automatic segmentation of these fascies is extremely hard to achieve. The ‘Medstation’ software was used to extract coordinates of muscle and tendon insertions by hand and combine them in a common frame of reference. CONCLUSIONS: A procedure has been established to extract the geometrical data of muscles, tendons and osseous structures important for the biomechanical model. For this model, extended muscle and tendon insertions have to be reduced to a point by calculation of the center of mass of the insertion area. A table for the positioning of the probationer enabled positioning with reproducible joint angles under exertion of strain. To define the relative position of different scans a screen of markers was integrated into this plate.