Breathing Thorax Model From CT images for use in Radiotherapy

Jef Vandemeulebroucke1,2,3,Jan Kybic1, David Sarrut2,3 , Patrick Clarysse2,

1 CMP, Department of Cybernetics, Czech Technical University in Prague, Prague, Czech Republic
2 CREATIS, Institut National des Sciences Appliquées de Lyon, Lyon, France
2 Leon Berard Cancer Center, Lyon, France
 

Motivation

External beam radiotherapy consists of exposing tumor cells to ionizing radiation. It is the primary treatment modality for non-operable lung cancer. The entire treatment is planned on CT images:

(a) CT image of the thorax with tumor present in the right lung

(b) Target volume delineation

(c) Treatment planning through simulation of the radiation
Figure 1 - Illustration of the planning process for radiotherapy
 

Breathing motion causes uncertainties during treatment planning and treatment delivery.

(a) It causes image artifacts and reduced image quality

(b) It has to be accounted for during planning, e.g. by applying margins around the tumor volume.
Figure 2 - Influence of respiratory motion on image quality and treatment planning

Proposed Methods

Two recently developped CT acquisition techniques have become widely available in radiotherapy. A 4DCT image (Figure 3a) consist of several 3DCT images at different breathing phases and can be used for treatment planning. Cone-beam CT ussually consists of one 3D image reconstructed from a large number of wide angle projections (Figure 3b). These images are acquired in the treatment room, moments before treatment, and can be used to verify the patient set-up. We wish to combine a 4DCT image and the cone-beam CT projection data of the same patient to make a patient specific predictive motion model

(a)

(b)
Figure 3 - A 4DCT image (a) and part of a cone-beam projection sequence (b) of the same patient

Schematically we can illustrate the necessary steps as in Figure 4:

  • Estimate motion in 4DCT using deformable registration. We use this result to build a patient-specific motion model
  • The prior model is used to estimate the motion present in a cone-beam projection sequence.
  • During cone-beam acquisition we can also acquire external signals like spirometry or abdominal height. We can use them to predict the internal motion.
  • The predictive model can be used during treatment. In combination with the external signals we can take the breathing into account during treatment delivery.

Figure 4

Publications