cs.CVOct 22, 2026

Diffusion-Based Synthesis of Complete 3D Biparametric Prostate MRI

Authors: Claudia Giardina, Montse Pardàs, Verónica Vilaplana

Organizations: Universitat Politècnica de Catalunya, Barcelona, Spain

Read PDF
Ask AI
Cite

Copy citation

DOINot available for this paper

Abstract

Synthetic biparametric MRI (bpMRI) may help alleviate the limited availability of large and representative prostate MRI datasets for deep learning applications. In this work, we present a conditional 3D diffusion framework for anatomy-conditioned synthesis of complete prostate bpMRI from anatomical masks. The framework supports sequential synthesis, in which T2w and diffusion modalities are generated by separately trained models, and joint synthesis, in which all bpMRI modalities are generated simultaneously within a single model. Five configurations were trained using PI-CAI, FastMRI Prostate, and PROSTATEx datasets to investigate different conditioning schemes, target modalities, and synthesis strategies. Evaluation using distribution-based metrics, paired-image similarity, and qualitative assessment demonstrated the feasibility of both synthesis pathways. Joint synthesis achieved competitive distributional fidelity while generating complete bpMRI, whereas direct synthesis of ADC and computed high-b-value DWI outperformed their derivation from synthetic diffusion-weighted images.

Figures & tables

Figure 1: Overview of the proposed framework for anatomy-conditioned prostate bpMRI synthesis. (a) Architecture of the conditional 3D diffusion model with anisotropy-aware depth regulation. (b) Example of the condition-guided reverse diffusion process for mask-conditioned T2w synthesis. The same formulation supports alternative conditioning and target configurations.
TargetsStrategyCondition →\rightarrow target(s)
LBV, HBVSequentialM1: Mask →\rightarrow T2w
  M2: Mask + T2w →\rightarrow LBV, HBV
 JointM4: Mask →\rightarrow T2w, LBV, HBV
ADC, cHBVSequentialM1: Mask →\rightarrow T2w
  M3: Mask + T2w →\rightarrow ADC, cHBV
 JointM5: Mask →\rightarrow T2w, ADC, cHBV
Table 1: Experimental configurations for sequential and joint prostate bpMRI synthesis. M1–M5 denote the five trained model configurations. In sequential synthesis, models are listed in their order of application.
MetricM1M4M5
FID ↓\downarrow53.6665.6473.02
KID ↓\downarrow0.050.060.08
Precision ↑\uparrow0.410.650.59
Recall ↑\uparrow0.070.090.09
Table 2: Distribution-based metrics for T2w synthesis.
Figure 2: Representative T2w synthesis examples from M1, M4, and M5 for two augmented masks.
LBVHBV
MetricM2M4MetricM2M4
FID ↓\downarrow32.4923.48FID ↓\downarrow37.5725.50
KID ↓\downarrow0.0210.016KID ↓\downarrow0.0240.016
Prec. ↑\uparrow0.790.86Prec. ↑\uparrow0.460.67
Recall ↑\uparrow0.400.49Recall ↑\uparrow0.200.18
Table 3: Distribution-based metrics for diffusion modality synthesis. For ADC and cHBV, the column “M4-derived” corresponds to modalities computed from synthetic LBV and HBV generated by M4.
Figure 3: Representative LBV/HBV synthesis examples from the sequential model M2 (left) and the joint model M4 (right).
Figure 4: Representative ADC/cHBV synthesis examples from the sequential model M3 (left) and the joint model M5 (right).
Figure 5: Comparison of derived (left) and directly generated (right) ADC and cHBV. The lesion is highlighted by the red circle.
More in this topic

Medical Image Synthesis

Medical image synthesis creates, translates, reconstructs, and edits clinical images to fill missing data and support analysis.

See all 310 papers

Explore similar work

CardsList

  1. WFM: 3D Wavelet Flow Matching for Ultrafast Multi-Modal MRI Synthesis

    Apr 22, 2026Yalcin Tur, Mihajlo Stojkovic, Ulas BagciFlow MatchingBraTS

    Department of Computer Science, Stanford University, Stanford, CA, USA · Machine and Hybrid Intelligence Lab, Feinberg School of Medicine, Northwestern University, Chicago, IL, USA

  2. Wavelet-Fusion Diffusion Model for Multimodal Brain MRI Synthesis with Modality and Metadata Conditioning

    May 30, 2026Muhammad Nabi Yasinzai, Remika Mito, Mangor PedersenVariational AutoencodersLatent Diffusion Models

    Department of Psychology & Neuroscience, Auckland University of Technology, Auckland, New Zealand. · Department of Psychiatry, University of Melbourne, Melbourne, Victoria 3053, Australia. · Auckland University of Technology, Auckland, New Zealand.

  3. Exploiting Completeness Perception with Diffusion Transformer for Unified 3D MRI Synthesis

    Feb 20, 2026Junkai Liu, Nay Aung, Theodoros N. Arvanitis +3Medical Image Synthesis3D Medical Imaging

    School of Engineering, University of Birmingham, UK · William Harvey Research Institute, Queen Mary University London, UK · Barts Heart Centre, St Bartholomew’s Hospital, Barts Health NHS Trust, UK +1