Multi-label vs multi-class classification of blood cells in microfluidic channels

A new study examines how to classify blood cells and their aggregates imaged by deformability cytometry, a high-throughput imaging flow cytometry technique that measures cellular stiffness alongside properties like area and elongation. The researchers compare multi-label and multi-class machine learning approaches for identifying individual cells and clusters within microfluidic channels. The work aims to determine which labeling scheme better captures cases where multiple cell types or aggregates appear in a single measurement.