Reconstruct or Generate: Exploring the Spectrum of Generative Modeling for Cardiac MRI
This work addresses the need to understand model trade-offs for medical imaging practitioners, but it is incremental as it benchmarks existing methods without introducing new ones.
The paper tackled the problem of comparing generative models for cardiac MRI tasks, finding that diffusion models excel in perceptual quality for unconditional generation but hallucinate with high masking ratios, while autoregressive models maintain stable perceptual performance across masking levels but with lower fidelity.
In medical imaging, generative models are increasingly relied upon for two distinct but equally critical tasks: reconstruction, where the goal is to restore medical imaging (usually inverse problems like inpainting or superresolution), and generation, where synthetic data is created to augment datasets or carry out counterfactual analysis. Despite shared architecture and learning frameworks, they prioritize different goals: generation seeks high perceptual quality and diversity, while reconstruction focuses on data fidelity and faithfulness. In this work, we introduce a "generative model zoo" and systematically analyze how modern latent diffusion models and autoregressive models navigate the reconstruction-generation spectrum. We benchmark a suite of generative models across representative cardiac medical imaging tasks, focusing on image inpainting with varying masking ratios and sampling strategies, as well as unconditional image generation. Our findings show that diffusion models offer superior perceptual quality for unconditional generation but tend to hallucinate as masking ratios increase, whereas autoregressive models maintain stable perceptual performance across masking levels, albeit with generally lower fidelity.