A continuous 4D motion model from multiple respiratory cycles for use in lung radiotherapy

Area of Science:

• Medical Physics
• Radiotherapy
• Image-guided therapy

Background:

• Respiratory motion during lung cancer radiotherapy introduces significant planning and delivery errors.
• Four-dimensional computed tomography (4DCT) is used to mitigate these errors, but datasets are memory-intensive.
• Existing methods may not fully account for variations in individual respiratory cycles.

Purpose of the Study:

• To develop and validate a novel computational method for constructing motion models from 4DCT data.
• To create models that represent an average respiratory cycle, reducing the impact of cycle-to-cycle variations.
• To enable automatic target propagation and dose summation across the respiratory cycle.

Main Methods:

• Constructing computational motion models from free-breathing CT (FBCT) data acquired in cine mode.
• Nonrigidly registering a reference breath-hold CT volume to multiple FBCT volumes.
• Temporally fitting registration parameters using B-splines to model an average respiratory cycle.
• Concatenating slab models to predict motion over the entire region of interest.

Main Results:

• Motion models demonstrated high accuracy with a mean target registration error (TRE) of 1.3 mm.
• The mean target model error (TME) was 1.6 mm, indicating precise prediction of anatomical point locations.
• The mean continuity error at slab boundaries was 2.2 mm, comparable to CT slice thickness (1.5 mm).

Conclusions:

• The developed computational motion models accurately represent average respiratory motion.
• These models offer a memory-efficient alternative to full 4DCT datasets.
• The models facilitate improved accuracy in radiotherapy planning and delivery for lung cancer patients.