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

2025-05-06

AI Summary

To address the scarcity of cardiac MRI (CMR) data and insufficient coverage of healthy phenotypes—key bottlenecks hindering AI pretraining and clinical deployment—this paper proposes the Phenotype-Guided Generative Framework (CPGG), a two-stage method for synthesizing high-fidelity, multi-phenotype dynamic CMR sequences. CPGG introduces a novel Masked Autoregressive Diffusion model (MAR-Diffusion) driven by clinically interpretable cardiac phenotypes, uniquely integrating anatomical constraints with spatiotemporal motion modeling as generative priors. It employs a GAN-based pretraining stage guided by a phenotype encoder, followed by conditional diffusion fine-tuning. Evaluated across multiple public and private CMR datasets, CPGG improves average accuracy in downstream diagnostic classification and phenotype prediction by 12.3%, expands effective pretraining data volume by over an order of magnitude, and significantly enhances model generalizability and clinical interpretability.

Technology Category

• Computer Vision: Diffusion Models for Vision
• Natural Language Processing: Code Generation / Program Synthesis from Natural Language
• Machine Learning: Deep Generative Models & Autoencoders

Application Category

• Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Web Mining and Content Analysis: Large pretrained models with web data

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.

Problem

• Limited availability of high-quality cardiac MRI datasets
• Insufficient unlabeled data for AI model pretraining
• Need for diverse synthetic cardiac MRI data generation

Innovation

• Phenotype-guided generative model for CMR synthesis
• Two-stage framework with masked autoregressive diffusion
• Generates diverse high-fidelity cardiac MRI data