Computing Sickle Erythrocyte Health Index on quantitative phase imaging and machine learning

Area of Science:

• Biomedical Engineering
• Hematology
• Medical Diagnostics

Background:

• Sickle cell disease (SCD) is a genetic blood disorder affecting red blood cells (RBCs), leading to severe health complications and reduced lifespan.
• Current diagnostic methods for SCD may not fully capture the dynamic health status of erythrocytes, necessitating advanced assessment tools.

Purpose of the Study:

• To develop and validate an in vitro quantitative assessment, the Sickle Erythrocyte Health Index (SEHI), for evaluating red blood cell health in SCD.
• To utilize quantitative phase imaging (QPI) and machine learning (ML) for a comprehensive erythrocyte health model.

Main Methods:

• Quantitative phase imaging (QPI) was employed to capture high-resolution images of red blood cells.
• Machine learning models, including deep learning, were used to analyze cell deformation, sickle-shape classification, and membrane flexibility.
• The SEHI was calculated using both hand-crafted and learned features from QPI data.

Main Results:

• The SEHI demonstrated statistically significant differences in erythrocyte health between individuals with and without SCD.
• Exposure to sodium metabisulfite induced increased sickling and decreased membrane flexibility in SCD blood samples, which was detected by the SEHI.
• Both hand-crafted and deep learning-based SEHI calculations proved sensitive to these induced changes.

Conclusions:

• The developed Sickle Erythrocyte Health Index provides a robust in vitro measure of red blood cell health in SCD.
• The SEHI holds potential clinical implications for improved SCD management and treatment decision-making.
• Further validation in diverse patient populations is recommended to establish broad clinical utility.