Track papers across labs, topics, and trends.
Search papers, labs, and topics across Lattice.
This study investigates the factors contributing to cross-dataset generalization failures in echocardiographic left ventricular segmentation by analyzing handcrafted ultrasound descriptors, VAE latent features, and segmentation-derived latent features across six datasets. The findings reveal that geometry-aware preprocessing can significantly mitigate transfer degradation, indicating that inconsistencies in field-of-view and framing are primary contributors to domain shift rather than intrinsic acoustic differences. Notably, the research establishes a robust predictive model for assessing absolute Dice drop on source-target pairs, achieving an R-squared value of 0.612, thereby providing a framework for monitoring transfer risk in clinical applications.
Geometry-aware preprocessing can substantially reduce echocardiographic segmentation errors caused by domain shifts, revealing that framing inconsistencies are more critical than acoustic differences.
Cross-dataset generalisation remains a major barrier to clinical deployment of echocardiographic left ventricular segmentation, yet the sources of this shift are rarely disentangled. We examined whether transfer degradation could be estimated before deployment using handcrafted ultrasound descriptors, VAE latent features, and segmentation-derived latent features across six echocardiographic datasets. Geometry-aware preprocessing substantially improved several poor transfer cases, suggesting that much of the apparent domain shift reflects field-of-view and framing inconsistencies rather than intrinsic acoustic differences alone. Intensity z-normalisation changed dataset separability by less than 0.005, indicating that brightness and contrast are not the dominant shift axis. Absolute Dice drop on held-out source-target pairs was predicted with an R-squared value of 0.612, an MAE of 0.082, and a Spearman rho of 0.681. The variant without LV and fan-shaped features retained approximately 70% of this explanatory power, supporting mask-free transfer-risk monitoring. The most informative discrepancy measure depended on the representation, with CMD strongest in z-normalised handcrafted features, with an absolute r of approximately 0.86 and an R-squared value of approximately 0.70; log-Wasserstein strongest in VAE space, with an r of approximately -0.90 and an R-squared value of approximately 0.81; and log-MMD strongest in LV-segmentation latent features, with an r of approximately -0.92 and an R-squared value of approximately 0.84. Apparent vendor effects were largely dataset-confounded. Echocardiographic domain shift is therefore structured and measurable, and its impact on segmentation can be partly reduced through geometry-aware preprocessing and anticipated using representation-specific transfer-risk estimation.