Hersh Sagreiya, MD

COVID-19 is a global pandemic with far-reaching consequences, and portable ultrasound is increasingly playing a key role in its assessment. We developed the Calculated Lung Ultrasound (CLU) technique, which automatically quantifies the degree of lung involvement, allowing evaluation of disease progression and response to treatment. It includes a prognostic score, the CLU index. We seek to further develop the CLU technique, incorporate more clinical data into its prediction, correlate the CLU score with patient outcomes, evaluate its robustness and reproducibility on a large dataset, and extend it to other disease, such as pneumonia and COPD.

Update:

We further developed a software platform and tool—Calculated Lung Ultrasound (CLU)—that uses machine learning via an unsupervised learning technique to automatically quantify lung ultrasound findings related to COVID-19, without requiring manual annotation. We published our results applying this algorithm to a multi-institutional dataset in the Diagnostics journal. The algorithm was refined to facilitate its use in different clinical environments worldwide and across different devices. We carefully curated a dataset at the University of Pennsylvania with 472 ultrasound movie clips with their corresponding clinical diagnosis. This dataset is now ready for analysis with CLU and other machine learning tools.