Work

Early detection and real-time monitoring of thrombogenesis in extracorporeal circuits through large-scale image-based profiling of circulating platelets

Year: 2026

Type: article

Abstract: Extracorporeal membrane oxygenation (ECMO) provides life-saving support for patients with severe cardiac or respiratory failure, yet intra-circuit thrombosis remains a frequent and serious complication. Despite anticoagulation, thrombosis occurs in up to 35% of patients, impairing device performance, increasing morbidity and mortality, and often necessitating urgent circuit replacement. Current monitoring methods rely on visual inspection or indirect assays, which are insufficiently sensitive for early thrombus detection. Here we present a direct, quantitative, real-time strategy for large-scale image-based analysis of circulating platelets to monitor intra-circuit thrombogenesis during ECMO. Using an optofluidic time-stretch microscope with 780 nm spatial resolution and throughput exceeding 10, 000 events per second, we imaged circulating platelets in Capra hircus (goats) as a large animal model for long-term ECMO and analyzed them with deep learning. This enabled continuous tracking of single platelets, platelet–platelet aggregates, and platelet–leukocyte aggregates. Significant increases in the concentration of circulating platelet aggregates consistently preceded pump exchange events ( p = 0.0026), confirming their value as an early indicator of intra-circuit thrombosis. The method also distinguished thrombotic from inflammatory events, as only platelet–leukocyte aggregates rose during systemic inflammation, and aggregate levels correlated with lactate dehydrogenase, linking thrombosis to hemolysis or tissue injury. These findings demonstrate that high-throughput, image-based platelet phenotyping provides a sensitive, real-time tool for monitoring ECMO thrombosis, enabling early risk detection and supporting optimized anticoagulant management.

Source: Journal of Physics Photonics

DOI: 10.1088/2515-7647/ae3930