Wahle A, Wellnhofer
E, Mugaragu I, Beier J, Oswald H, Fleck E:
CAR '93, in:
Proceedings of the 7th International Symposium on Computer Assisted Radiology, Berlin, Springer
Page 792, 1993
Enhancements of the imaging geometry assessments by using biplane calibration methods (Links)
Abstract: 3-D reconstruction of thoracic vessels aids in clinical management of patients with any kind of heart disease. Two simultaneous - but not necessarily orthogonal - angiographic views give 2-D projections of the 3-D space. Based upon known imaging geometry a 3-D model of complex vessel structures for visualization and quantification can be reconstructed.
The quantification quality strongly depends on the accuracy and completeness of the imaging parameters. To calculate missing and to correct distorted parameters a new geometry approximation has been developed. The initial geometry is based on the geometric parameters of the X-ray device. This status is improved using several corresponding vessel points in both projections. The reconstruction errors of these points are analysed and correction coefficients are calculated. This is carried out in an iterative process.
To achieve accurate measurements in absolute dimensions a new calibration procedure was realized. The size of the image intensifiers can be determined by lead markers mounted on the entrance fields. Since continuous phantom measurements are impracticable and catheter radius determinations are prone to gross errors, an object of known dimensions and sufficient size visible in both projections was searched for. Markers on a catheter or the maximum diameter of the loop of a pigtail catheter were used as reference objects. The relation of actual to reconstructed object size provides additional correction coefficients for the approximation process.
The biplane calibration allows a reliable evaluation without phantom measurements or any assumptions. Common problems such as parallaxes do not occur. In addition with further corrections (image distortion, pointspreading) an accurate quantitative 3-D reconstruction is performed. The algorithm was validated using different phantoms of known sizes.