Lung Ultrasound Segmentation and Adaptation between COVID-19 and Community-Acquired Pneumonia

TL;DR

This study develops deep learning methods for lung ultrasound B-line segmentation to distinguish COVID-19 from other pneumonia, demonstrating that domain adaptation improves accuracy with limited data, aiding clinical diagnosis during epidemics.

Contribution

The paper introduces a domain adaptation approach for lung ultrasound segmentation between COVID-19 and CAP, showing significant performance improvements with minimal data in a real-world setting.

Findings

Domain adaptation improved Dice scores from 0.60 to 0.87 for COVID-19 cases.

Adapting from COVID-19 to CAP did not improve performance.

Performance correlates with label consistency and data domain diversity.

Abstract

Lung ultrasound imaging has been shown effective in detecting typical patterns for interstitial pneumonia, as a point-of-care tool for both patients with COVID-19 and other community-acquired pneumonia (CAP). In this work, we focus on the hyperechoic B-line segmentation task. Using deep neural networks, we automatically outline the regions that are indicative of pathology-sensitive artifacts and their associated sonographic patterns. With a real-world data-scarce scenario, we investigate approaches to utilize both COVID-19 and CAP lung ultrasound data to train the networks; comparing fine-tuning and unsupervised domain adaptation. Segmenting either type of lung condition at inference may support a range of clinical applications during evolving epidemic stages, but also demonstrates value in resource-constrained clinical scenarios. Adapting real clinical data acquired from COVID-19 patients to those from CAP patients significantly improved Dice scores from 0.60 to 0.87 (p < 0.001) and from 0.43 to 0.71 (p < 0.001), on independent COVID-19 and CAP test cases, respectively. It is of practical value that the improvement was demonstrated with only a small amount of data in both training and adaptation data sets, a common constraint for deploying machine learning models in clinical practice. Interestingly, we also report that the inverse adaptation, from labelled CAP data to unlabeled COVID-19 data, did not demonstrate an improvement when tested on either condition. Furthermore, we offer a possible explanation that correlates the segmentation performance to label consistency and data domain diversity in this point-of-care lung ultrasound application.