3D-LLDM: Label-Guided 3D Latent Diffusion Model for Improving High-Resolution Synthetic MR Imaging in Hepatic Structure Segmentation

TL;DR

This paper introduces 3D-LLDM, a label-guided 3D latent diffusion model that generates high-quality synthetic MR images with segmentation masks, enhancing data availability and improving tumor segmentation accuracy.

Contribution

The paper presents a novel 3D latent diffusion model guided by anatomical labels, specifically designed for high-resolution synthetic MR image generation in hepatic imaging.

Findings

Achieves a Fréchet Inception Distance of 28.31, outperforming GANs and diffusion baselines.

Synthetic data improves hepatocellular carcinoma segmentation by up to 11.15% Dice score.

Demonstrates effective use of label-guided volumetric synthesis for medical imaging.

Abstract

Deep learning and generative models are advancing rapidly, with synthetic data increasingly being integrated into training pipelines for downstream analysis tasks. However, in medical imaging, their adoption remains constrained by the scarcity of reliable annotated datasets. To address this limitation, we propose 3D-LLDM, a label-guided 3D latent diffusion model that generates high-quality synthetic magnetic resonance (MR) volumes with corresponding anatomical segmentation masks. Our approach uses hepatobiliary phase MR images enhanced with the Gd-EOB-DTPA contrast agent to derive structural masks for the liver, portal vein, hepatic vein, and hepatocellular carcinoma, which then guide volumetric synthesis through a ControlNet-based architecture. Trained on 720 real clinical hepatobiliary phase MR scans from Samsung Medical Center, 3D-LLDM achieves a Fr\'echet Inception Distance (FID) of 28.31, improving over GANs by 70.9% and over state-of-the-art diffusion baselines by 26.7%. When used for data augmentation, the synthetic volumes improve hepatocellular carcinoma segmentation by up to 11.153% Dice score across five CNN architectures.