PIVM: Diffusion-Based Prior-Integrated Variation Modeling for Anatomically Precise Abdominal CT Synthesis

2026-03-23

AI Summary

Abdominal CT imaging suffers from data scarcity due to high annotation costs and privacy constraints, which hinders the development of segmentation and diagnostic models. To address this challenge, this work proposes the PIVM framework, which uniquely integrates organ-specific intensity priors with segmentation labels directly into the diffusion process to synthesize CT images in the native image space. The generated images preserve the full Hounsfield Unit (HU) range, exhibit sharp anatomical boundaries, and retain fine textural details. By modeling voxel-wise intensity deviations from the learned priors, PIVM achieves precise anatomical fidelity and effectively mitigates the blurriness commonly observed in conventional generative approaches. This advancement significantly enhances performance on downstream medical analysis tasks.

Technology Category

• Computer Vision: Diffusion Models for Vision
• Machine Learning: Deep Generative Models & Autoencoders
• Natural Language Processing: Generation

Application Category

• User Modeling, Personalization and Recommendation: User privacy protection in personalized systems
• Security and Privacy: Data transparency and provenance
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs

Abstract

Abdominal CT data are limited by high annotation costs and privacy constraints, which hinder the development of robust segmentation and diagnostic models. We present a Prior-Integrated Variation Modeling (PIVM) framework, a diffusion-based method for anatomically accurate CT image synthesis. Instead of generating full images from noise, PIVM predicts voxel-wise intensity variations relative to organ-specific intensity priors derived from segmentation labels. These priors and labels jointly guide the diffusion process, ensuring spatial alignment and realistic organ boundaries. Unlike latent-space diffusion models, our approach operates directly in image space while preserving the full Hounsfield Unit (HU) range, capturing fine anatomical textures without smoothing. Source code is available at https://github.com/BZNR3/PIVM.

Problem

• abdominal CT synthesis
• data scarcity
• annotation cost
• privacy constraints
• medical image generation

Innovation

• diffusion model
• anatomically precise synthesis
• intensity prior
• image-space generation
• abdominal CT