Non-Invasive White Blood Cell Screening

Abstract

Neutropenia, an abnormally low count of the most abundant type of white blood cells (neutrophils), is a common side effect of chemotherapy treatments, and a frequent cause of serious infection episodes. Current methods to measure neutropenia require blood extraction and do not allow for frequent testing. In this work, a non-invasive method that can frequently and automatically screen patients for severe neutropenia is proposed. This method involves the acquisition of one-minute microscopy videos of the microcirculation through the skin in finger-nailfold capillaries, followed by image analysis and binary classification of severe neutropenia by an automated software pipeline. We validated this method in a study with chemotherapy patients who underwent autologous stem cell transplantation. The results demonstrate that our automated method accurately classifies severely neutropenic cases in 35 patients with an area under the curve (AUC) of 0.95.