Learning Generic Lung Ultrasound Biomarkers for Decoupling Feature Extraction from Downstream Tasks

TL;DR

This paper introduces a method to learn interpretable lung ultrasound biomarkers through a pre-task, enabling flexible and cost-effective feature extraction that supports various clinical tasks with comparable accuracy to end-to-end models.

Contribution

It proposes a decoupled feature learning approach using biomarker classification, reducing training costs and enhancing interpretability for lung ultrasound analysis.

Findings

Biomarker features are effective for multiple clinical tasks.

Models trained on weak supervision achieve comparable accuracy to end-to-end models.

Decoupled features reduce training costs significantly.

Abstract

Contemporary artificial neural networks (ANN) are trained end-to-end, jointly learning both features and classifiers for the task of interest. Though enormously effective, this paradigm imposes significant costs in assembling annotated task-specific datasets and training large-scale networks. We propose to decouple feature learning from downstream lung ultrasound tasks by introducing an auxiliary pre-task of visual biomarker classification. We demonstrate that one can learn an informative, concise, and interpretable feature space from ultrasound videos by training models for predicting biomarker labels. Notably, biomarker feature extractors can be trained from data annotated with weak video-scale supervision. These features can be used by a variety of downstream Expert models targeted for diverse clinical tasks (Diagnosis, lung severity, S/F ratio). Crucially, task-specific expert models are comparable in accuracy to end-to-end models directly trained for such target tasks, while being significantly lower cost to train.