PhD thesis of Minh Hung Nguyen on 07/23/21

PhD thesis of Minh Hung Nguyen from TIMC GMCAO team will take place on July the 23th at 2pm:

« Data Consistency Conditions in 3D Tomography
and Scanner Calibration using Analytic Approaches »

       (Defense in english)

bullet  Thesis supervision:

  • Laurent Desbat, Professeur, Université Grenoble Alpes, laboratoire TIMC UMR 5525
  • Rolf Clackdoyle, Directeur de Recherche, CNRS, laboratoire TIMC UMR 5525

bullet  Jury:

  • Michel Defrise, Professeur, University Hospital Az-Vub, Reporter
  • Xiaochuan Pan, Professeur, University Of Chicago, Reporter
  • Valérie Perrier, Professeure Des Universités, Grenoble Inp, Examiner
  • Françoise Peyrin, Directrice De Recherche, INSERM, Délégation Auvergne-Rhône-Alpes, Examiner
  • Charles  Soussen, Professeur Des Universités, CentraleSupélec, Examiner
  • Yannick Grondin, Ingénieur Docteur, Surgiqual Institute, Examiner

bullet  Key words:  

range conditions, geometric calibration, data consistency conditions, tomography, cone-beam geometry

bullet  Abstract:

This work concerns the data consistency conditions and their applications in geometric self-calibration. In Medical Imaging, an object is projected through a mechanical system and the corresponding projections must satisfy certain conditions if the system is consistent. These conditions are called data consistency conditions (DCC). In the situation that the object is assumed to be unknown, DCC play an important role to calibrate the geometric parameters of the system only from the projection data. Our work on one hand is to derive new DCC in different geometry contexts, and on the other hand is to try to appropriately apply them into some corresponding geometric calibration problems. We investigate three geometry contexts: 3D parallel geometry, cone-beam geometry with linear sources and general
cone-beam geometry. With 3D parallel geometry, we present a pair-wise DCC leading to an analytic formula to calibrate the projection’s viewing direction in general case, and a technique of converting the 3D calibration problem into many different 2D calibration problems in a particular degenerate case. In the cone-beam geometry with linear sources, we reuse the above technique and give a method to calibrate the corresponding source position of the projection based on fan-beam consistency condition. For the last contribution with general cone-beam geometry, we derive new DCC with general source trajectory and apply them into a cone-beam calibration problem with circular source trajectory, where the source position on the circular trajectory is the parameter being calibrated.