Key result
An end-to-end reciprocal detection and tracking model using adversarial training demonstrated superiority in localizing anatomical landmarks in echocardiography despite sparse annotations.
Why the study?
Automatic localization of echocardiographic landmarks is desirable to improve workflow and reduce interobserver variability, but training machine learning models is hindered by sparse manual annotations across cardiac cine series.
Does an end-to-end reciprocal detection and tracking model improve the localization of anatomical landmarks in echocardiography with sparse annotations?
Does an end-to-end reciprocal detection and tracking model improve the localization of anatomical landmarks in echocardiography with sparse annotations?
A novel reciprocal detection and tracking model enables automatic localization of anatomical landmarks in echocardiography using very few annotated frames.
Supports efficient annotation use for echocardiography landmark detection; leaves open clinical workflow integration and outcome validation.
Localization of anatomical landmarks to perform two-dimensional measurements in echocardiography is part of routine clinical workflow in cardiac disease diagnosis. Automatic localization of those landmarks is highly desirable to improve workflow and reduce interobserver variability. Training a machine learning framework to perform such localization is hindered given the sparse nature of gold standard labels; only few percent of cardiac cine series frames are normally manually labeled for clinical use. In this paper, we propose a new end-to-end reciprocal detection and tracking model that is specifically designed to handle the sparse nature of echocardiography labels. The model is trained using few annotated frames across the entire cardiac cine sequence to generate consistent detection and tracking of landmarks, and an adversarial training for the model is proposed to take advantage of these annotated frames. The superiority of the proposed reciprocal model is demonstrated using a series of experiments.
No takes yet. Share an insight, caveat, or question.
Lin et al. (2021) studied Cardiac disease diagnosis (echocardiography). End-to-end reciprocal detection and tracking model was evaluated on Consistent detection and tracking of landmarks. An end-to-end reciprocal detection and tracking model using adversarial training demonstrated superiority in localizing anatomical landmarks in echocardiography despite sparse annotations.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: