Protect the Brain When Treating the Heart: A Convolutional Neural Network for Detecting Emboli

TL;DR

This paper introduces a 2.5D U-Net based method for real-time detection and segmentation of gaseous microemboli in cardiac ultrasound images, aiding surgical monitoring.

Contribution

The novel approach combines a 2.5D U-Net architecture with real-time processing for accurate GME detection in ultrasound imaging.

Findings

Achieved high segmentation accuracy of GME in ultrasound data.

Enabled real-time GME quantification during cardiac procedures.

Integrated into surgical protocols for improved patient monitoring.

Abstract

Gaseous microemboli (GME) represent a common complication of cardiac structural interventions across both surgical and transcatheter approaches. Transthoracic cardiac ultrasound imaging represents a convenient methodology to visualize the presence of circulating GME. However, their detection and quantification are far from trivial due to operator-dependent view, high velocity, and objects with similar structure in the background. Here, we propose an approach based on a 2.5D U-Net architecture to segment GME in space-time connected data. Such an approach yields robust detection against the background and high segmentation accuracy while retaining real-time execution speed. These properties facilitated the integration of the proposed pipeline into patient-monitoring surgical protocols, providing the quantification of GME area over time.