Synthesizing 4D image phantoms spanning multiple respiratory cycles from heterogeneous data sources

Area of Science:

• Medical Imaging
• Computational Anatomy
• Algorithm Development

Background:

• Realistic four-dimensional (4D) image datasets with ground truth (GT) motion information are essential for developing and validating motion tracking algorithms.
• Existing datasets often lack sufficient duration, spanning only a single respiratory cycle, and are scarce for multi-cycle motion analysis.

Purpose of the Study:

• To present a reproducible pipeline for synthesizing anatomically semi-realistic 4D motion phantoms with GT motion information.
• To leverage heterogeneous and publicly available data sources for phantom generation.
• To provide a resource for robust development, validation, and benchmarking of image-based motion tracking algorithms.

Main Methods:

• Integration of 4D deformation vector fields (DVFs) from single-cycle datasets, static 3D images, and multi-cycle respiratory traces.
• Harmonization of heterogeneous inputs via DVF alignment, spatial gap completion, and respiratory trace-guided DVF interpolation to create continuous motion fields.
• Application of synthesized motion fields to static images to generate time-resolved 4D image sequences capturing multi-cycle respiratory motion.

Main Results:

• Generation of three distinct phantom datasets (4D CT and 4D MR) with multi-organ GT segmentations and/or GT DVFs.
• Quantitative validation showing physically plausible volumetric changes and significant correlation between synthesized and measured liver motion.
• High qualitative realism scores (4.2/5) from expert assessment and comparable motion tracking performance to clinical sequences.
• Public accessibility of generated phantoms, source code, and input data for reproducibility and customization.

Conclusions:

• The developed pipeline successfully synthesizes 4D image phantoms capturing physiological motion over multiple respiratory cycles.
• The openly shared resource addresses a critical gap, enabling more robust development and validation of motion tracking algorithms.
• This work facilitates advancements in medical imaging research by providing a standardized evaluation setting.