Low-complexity lung ultrasound video scoring by means of intensity projection-based video compression

Umair Khan, Sajjad Afrakhteh, Federico Mento, Gizem Mert, Andrea Smargiassi, Riccardo Inchingolo, Francesco Tursi, Veronica Narvena Macioce, Tiziano Perrone, Giovanni Iacca, Libertario Demi

Abstract: Highlights•We propose a novel method for LUS video-level scoring based on video compression.•We exploit intensity projection-based strategy for performing the video compression.•We develop a SA CNN model trained using a custom loss function for LUS data scoring.•The computational cost improved ¿100x while maintaining state of the art performance.