Identification of non-activated lymphocytes using three-dimensional refractive index tomography and machine learning

Area of Science:

• Biophysics
• Immunology
• Computational Biology

Background:

• Accurate identification of lymphocyte subsets is critical for understanding disease mechanisms.
• Existing lymphocyte identification methods often involve time-consuming labeling techniques that can alter cell function.
• There is a need for non-invasive, label-free methods for single-cell lymphocyte analysis.

Purpose of the Study:

• To present a novel method for label-free identification of non-activated lymphocyte cell types at the single-cell level.
• To utilize refractive index (RI) tomography combined with machine learning for quantitative cell property retrieval and classification.
• To establish a versatile tool for investigating lymphocyte roles in various human diseases.

Main Methods:

• Acquisition of three-dimensional RI maps of individual lymphocytes using RI tomography.
• Quantitative retrieval of morphological and biochemical properties from RI data.
• Development and optimization of a machine learning classification model (k-NN, k=4) using these quantitative properties.

Main Results:

• The study successfully identified three major lymphocyte types: B cells, CD4+ T cells, and CD8+ T cells.
• The developed k-NN (k=4) machine learning model achieved high sensitivity and specificity in cell type classification.
• Morphological and biochemical features derived from RI tomography were effective for discriminating lymphocyte subtypes.

Conclusions:

• The combination of RI tomography and machine learning provides a novel, label-free approach for single-cell lymphocyte identification.
• This method offers a versatile and potentially cost-effective alternative to traditional cell labeling techniques.
• The technique holds promise for advancing research in immunology, oncology, autoimmune diseases, and infectious diseases.