Echocardiography video synthesis from end diastolic semantic map via diffusion model

TL;DR

This paper introduces a novel diffusion model-based method for synthesizing echocardiography videos from semantic maps of the initial cardiac frame, enhancing realism and coherence in medical video generation.

Contribution

It extends existing diffusion models with semantic guidance and spatial normalization to improve cardiac video synthesis from limited datasets.

Findings

Outperforms standard diffusion models on CAMUS dataset

Achieves lower FID, FVD, and SSMI scores

Enhances realism and coherence in echocardiography video synthesis

Abstract

Denoising Diffusion Probabilistic Models (DDPMs) have demonstrated significant achievements in various image and video generation tasks, including the domain of medical imaging. However, generating echocardiography videos based on semantic anatomical information remains an unexplored area of research. This is mostly due to the constraints imposed by the currently available datasets, which lack sufficient scale and comprehensive frame-wise annotations for every cardiac cycle. This paper aims to tackle the aforementioned challenges by expanding upon existing video diffusion models for the purpose of cardiac video synthesis. More specifically, our focus lies in generating video using semantic maps of the initial frame during the cardiac cycle, commonly referred to as end diastole. To further improve the synthesis process, we integrate spatial adaptive normalization into multiscale feature maps. This enables the inclusion of semantic guidance during synthesis, resulting in enhanced realism and coherence of the resultant video sequences. Experiments are conducted on the CAMUS dataset, which is a highly used dataset in the field of echocardiography. Our model exhibits better performance compared to the standard diffusion technique in terms of multiple metrics, including FID, FVD, and SSMI.