Coronavirus disease 2019 patients prognostic stratification based on low complex lung ultrasound video compression

• DOI: 10.1121/10.0018617
• Publication Date: 2023-04-27T18:39:22Z
• Authors: Umair Khan; Sajjad Afrakhteh; Federico Mento; Andrea Smargiassi; Riccardo Inchingolo; Francesco Tursi; Veronica Narvena; Tiziano Perrone; Giovanni Iacca; Libertario Demi

In the last years, efforts have been made towards automating semi-quantitative analysis of lung ultrasound (LUS) data. To this end, several methods proposed with a focus on frame-level classification. However, no extensive work has done to evaluate LUS data directly at video level. This study proposes an effective compression and classification technique for assessing is based maximum, mean, minimum intensity projection (with respect temporal dimension) allows preserving hyper- hypo-echoic regions results in compressing down three frames, which are then classified using convolutional neural network (CNN). Results show that not only preserves visual artifacts appearance reduced data, but also achieves promising agreement 81.61% prognostic Conclusively, suggested method reduces amount frames needed assess 3. Note average videos consists few hundreds frames. At same time, state-of-the-art performance levels achieved, while significantly reducing computational cost.