Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Phenotype-Guided Generative Model for High-Fidelity Cardiac MRI Synthesis: Advancing Pretraining and Clinical Applications

Ziyu Li, Yujian Hu, Zhengyao Ding, Yiheng Mao, Haitao Li, Fan Yi, Hongkun Zhang, Zhengxing Huang

TL;DR

The paper tackles the data scarcity problem in cardiac MRI by introducing Cardiac Phenotype-Guided CMR Generation (CPGG), a two-stage approach that first models cardiac phenotypes with a VAE and then generates high-fidelity CMR cine conditioned on these phenotypes using a masked autoregressive diffusion model with a diffusion loss that replaces vector quantization. This framework enables large-scale, fine-grained CMR synthesis with improved control over structural and functional heart features, and shows superior generation quality and faster inference compared to existing baselines. The generated synthetic data, when mixed into pretraining and finetuning, yields consistent and sometimes substantial gains on downstream tasks such as disease classification and cardiac phenotypes prediction across UK Biobank datasets and a private CMDS dataset, demonstrating practical value for pretraining, augmentation, and clinical analysis. Overall, CP GG advances pretraining for CMR AI and offers a scalable path toward broader clinical deployment of AI in cardiology.

Abstract

Cardiac Magnetic Resonance (CMR) imaging is a vital non-invasive tool for diagnosing heart diseases and evaluating cardiac health. However, the limited availability of large-scale, high-quality CMR datasets poses a major challenge to the effective application of artificial intelligence (AI) in this domain. Even the amount of unlabeled data and the health status it covers are difficult to meet the needs of model pretraining, which hinders the performance of AI models on downstream tasks. In this study, we present Cardiac Phenotype-Guided CMR Generation (CPGG), a novel approach for generating diverse CMR data that covers a wide spectrum of cardiac health status. The CPGG framework consists of two stages: in the first stage, a generative model is trained using cardiac phenotypes derived from CMR data; in the second stage, a masked autoregressive diffusion model, conditioned on these phenotypes, generates high-fidelity CMR cine sequences that capture both structural and functional features of the heart in a fine-grained manner. We synthesized a massive amount of CMR to expand the pretraining data. Experimental results show that CPGG generates high-quality synthetic CMR data, significantly improving performance on various downstream tasks, including diagnosis and cardiac phenotypes prediction. These gains are demonstrated across both public and private datasets, highlighting the effectiveness of our approach. Code is availabel at https://anonymous.4open.science/r/CPGG.

Phenotype-Guided Generative Model for High-Fidelity Cardiac MRI Synthesis: Advancing Pretraining and Clinical Applications

TL;DR

The paper tackles the data scarcity problem in cardiac MRI by introducing Cardiac Phenotype-Guided CMR Generation (CPGG), a two-stage approach that first models cardiac phenotypes with a VAE and then generates high-fidelity CMR cine conditioned on these phenotypes using a masked autoregressive diffusion model with a diffusion loss that replaces vector quantization. This framework enables large-scale, fine-grained CMR synthesis with improved control over structural and functional heart features, and shows superior generation quality and faster inference compared to existing baselines. The generated synthetic data, when mixed into pretraining and finetuning, yields consistent and sometimes substantial gains on downstream tasks such as disease classification and cardiac phenotypes prediction across UK Biobank datasets and a private CMDS dataset, demonstrating practical value for pretraining, augmentation, and clinical analysis. Overall, CP GG advances pretraining for CMR AI and offers a scalable path toward broader clinical deployment of AI in cardiology.

Abstract

Cardiac Magnetic Resonance (CMR) imaging is a vital non-invasive tool for diagnosing heart diseases and evaluating cardiac health. However, the limited availability of large-scale, high-quality CMR datasets poses a major challenge to the effective application of artificial intelligence (AI) in this domain. Even the amount of unlabeled data and the health status it covers are difficult to meet the needs of model pretraining, which hinders the performance of AI models on downstream tasks. In this study, we present Cardiac Phenotype-Guided CMR Generation (CPGG), a novel approach for generating diverse CMR data that covers a wide spectrum of cardiac health status. The CPGG framework consists of two stages: in the first stage, a generative model is trained using cardiac phenotypes derived from CMR data; in the second stage, a masked autoregressive diffusion model, conditioned on these phenotypes, generates high-fidelity CMR cine sequences that capture both structural and functional features of the heart in a fine-grained manner. We synthesized a massive amount of CMR to expand the pretraining data. Experimental results show that CPGG generates high-quality synthetic CMR data, significantly improving performance on various downstream tasks, including diagnosis and cardiac phenotypes prediction. These gains are demonstrated across both public and private datasets, highlighting the effectiveness of our approach. Code is availabel at https://anonymous.4open.science/r/CPGG.
Paper Structure (9 sections, 3 equations, 3 figures, 3 tables)

This paper contains 9 sections, 3 equations, 3 figures, 3 tables.

Table of Contents

  1. Introduction
  2. Method
  3. Cardiac Phenotypes Generation
  4. Masked Autoregressive CMR Generative Model without Vector Quantization
  5. Experiments
  6. Datasets and Experiments Setting
  7. CMR Generation Quality
  8. Performance of Downstream Tasks after Data Mixing
  9. Conclusion

Figures (3)

  • Figure 1: Overview of our model. A and C describe a two-stage generation process. B showed the details of the generation of each token.
  • Figure 2: Examples of generated CMR and their corresponding cardiac phenpotypes using the CPGG framework, ordered by LVEDV from small to large.
  • Figure 3: Performance of disease classification task on private dataset.