Assisted diagnosis algorithm for lung ultrasound in COVID-19 patients

Lung ultrasound has become one of the most promising medical techniques for the diagnosis and monitoring of pneumonia, which is one of the main complication of SARS-CoV-2 infection. Despite this, the lack of trained personnel in lung echography has restricted its use worldwide. Computer aided diagnosis could help reducing the learning curve for less experienced technicians and, therefore, extending the use of lung ultrasound more quickly, while reducing the exam duration. This work explores the feasibility of real-time image processing algorithms for automatic calculation of the lung score. A clinical trial with 30 patients was completed following the same protocol of acquiring saving 3 seconds videos of different thorax zones. Those videos were evaluated by an experienced physician and by a custom developed algorithm for detecting A-lines, B-lines, and consolidations. The concordance between both findings were 88% for B-lines, 93.4% for consolidations and 70.2% for A-lines, reducing the acquisition time using the ULTRACOV prototype [1] by more than half compared to a conventional scanner. The good agreement of the results proves the feasibility of implementing real-time algorithms for aided diagnosis in lung ultrasound equipment. © 2022 IEEE.

RL Based Unsupervised Video Summarization Framework for Ultrasound Imaging

The need for summarizing long medical scan videos for automatic triage in Emergency Departments and transmission of the summarized videos for telemedicine has gained significance during the COVID-19 pandemic. However, supervised learning schemes for summarizing videos are infeasible as manual labeling of scans for large datasets is impractical by frontline clinicians. This work presents a methodology to summarize ultrasound videos using completely unsupervised learning schemes and is validated on Lung Ultrasound videos. A Convolutional Autoencoder and a Transformer decoder is trained in an unsupervised reinforcement learning setup i.e., without supervised labels in the whole workflow. Novel precision and recall computation for ultrasound videos is also presented employing which high Precision and F1 scores of 64.36% and 35.87% with an average video compression rate of 78% is obtained when validated against clinically annotated cases. Even though demonstrated using lung ultrasound videos, our approach can be readily extended to other imaging modalities. © 2022, The Author(s), under exclusive license to Springer Nature Switzerland AG.