Track papers across labs, topics, and trends.
Search papers, labs, and topics across Lattice.
AbdomenGen, a sequential volume-conditioned diffusion framework, is introduced for generating controllable abdominal anatomy phantoms. The framework uses a novel Volume Control Scalar (VCS) to decouple organ size from body habitus, enabling interpretable and independent volume modulation across 11 abdominal organs. Experiments demonstrate strong geometric fidelity (e.g., liver Dice of 0.83) and the utility of VCS-based selection in reducing distributional distance within a hepatomegaly cohort by 73.6%, showcasing the framework's potential for clinical simulation studies.
Finally, a way to generate clinically meaningful anatomical variations with calibrated, independent control over multiple abdominal organs.
Computational phantoms are widely used in medical imaging research, yet current systems to generate controlled, clinically meaningful anatomical variations remain limited. We present AbdomenGen, a sequential volume-conditioned diffusion framework for controllable abdominal anatomy generation. We introduce the \textbf{Volume Control Scalar (VCS)}, a standardized residual that decouples organ size from body habitus, enabling interpretable volume modulation. Organ masks are synthesized sequentially, conditioning on the body mask and previously generated structures to preserve global anatomical coherence while supporting independent, multi-organ control. Across 11 abdominal organs, the proposed framework achieves strong geometric fidelity (e.g., liver dice $0.83 \pm 0.05$), stable single-organ calibration over $[-3,+3]$ VCS, and disentangled multi-organ modulation. To showcase clinical utility with a hepatomegaly cohort selected from MERLIN, Wasserstein-based VCS selection reduces distributional distance of training data by 73.6\% . These results demonstrate calibrated, distribution-aware anatomical generation suitable for controllable abdominal phantom construction and simulation studies.