Deformable vector fields warping for modelling of irregular breathing^∗

Area of Science:

• Medical Imaging
• Radiation Oncology
• Computational Biology

Background:

• 4D computed tomography (4DCT) is standard for imaging organ motion in radiotherapy but has limitations in capturing breathing variability.
• Accurate imaging of respiratory motion is crucial for effective radiotherapy planning, especially for lung cancer patients.

Purpose of the Study:

• To develop and validate a method for transferring breathing motion across longitudinal imaging datasets to incorporate intra-patient variability.
• To improve the definition of treatment volumes and margins in radiation therapy by better understanding patient-specific breathing motion.

Main Methods:

• Combined five repeated 4DCT scans from 6 non-small cell lung cancer patients into multi-breath datasets (m4DCT) using deformable image registration.
• Quantified intra-patient differences by evaluating tumor center of mass displacement and volume changes.
• Compared internal target volumes (ITVs) defined on m4DCT with those from conventional 4DCT.

Main Results:

• The method successfully merged repeated imaging into a continuum, showing no discontinuity between successive breaths.
• Tumor motion primarily occurred in the superior-inferior direction, with variability ranging from 14.4 mm to 0.1 mm depending on tumor location.
• Tumor volume exhibited significant expansion (up to 65%) and contraction (up to 74%) during inhalation and exhalation phases, potentially enlarging the ITV by up to 8%.

Conclusions:

• 4DCT can be extended to model variable breathing motion by synthesizing additional phases from multiple time-resolved images.
• Incorporating improved knowledge of patient breathing variability allows for a more precise definition of treatment volumes and margins in radiation therapy.