Search papers, labs, and topics across Lattice.
This paper addresses the challenge of limited realism in simulated echocardiographic data for training deep learning models for myocardial strain estimation. They introduce a novel simulation strategy incorporating speckle decorrelation measures from real videos and an iterative refinement process to enhance motion realism. The resulting open-source dataset of 1,478 videos enables training a motion estimation algorithm that outperforms clinical standards, achieving a GLS variability of 1.42% compared to 1.78% for clinical reference.
Synthesizing more realistic echocardiographic data lets deep learning models beat the clinical standard for myocardial strain estimation, a key biomarker for early diagnosis.
Speckle tracking echocardiography (STE) is the clinical standard for myocardial strain estimation. Despite good performance on global strain (GLS), its accuracy for regional strain remains limited, even though this biomarker is highly relevant for early diagnosis and the characterization of subtle abnormalities. from clinical data. Deep learning is a promising alternative, but its development is constrained by the lack of reliable motion references. Existing solutions rely either on STE-derived labels or on simulations generated by physics-based models, but these synthetic sequences still have limited realism compared with clinical data.In this paper, we propose a novel simulation strategy that incorporates speckle decorrelation measures from real videos and uses an iterative refinement process to improve the motion realism in the simulations. We created an open-source photorealistic dataset of 1,478 videos with reference motion, which was used to train an echocardiographic motion estimation algorithm. The proposed method achieves unmatched performance on global and regional strain, notably reaching a GLS variability of 1.42% in an inter-expert setting compared to 1.78% for the clinical reference.
