Identification of lymphoma types using 2D light scattering microscopy and machine learning

Abstract

Lymphomas encompass Hodgkin lymphoma (HL) and non-Hodgkin lymphoma (NHL). Accurate clinical diagnosis is paramount for informed treatment and prognosis, and lymphoma classification plays a central role in guiding treatment strategies and evaluating treatment outcomes. In this study, we have developed a static cytometry leveraging laser and microscope technology to capture two-dimensional (2D) light scattering patterns of individual cells. Within this method, a single lymphoma cell is positioned in a liquid-based chip and vertically stimulated by a 532 nm green laser. The resulting light scattering pattern of the cell is observed and recorded by a COMS detector through a microscope optical system, covering a polar angle range of 75 to 105 degrees. The light scattering pattern exhibited by lymphoma cells displays distinct speckles, with the texture features of these speckles influenced by internal cell structures, including organelle count and their spatial arrangement. By extracting and analyzing the characteristic values from these cell scattering patterns, we can achieve lymphoma cell identification. In this study, we successfully differentiated between HDLM-2 cells (HL) and Daudi cells (NHL) using the machine learning support vector machine (SVM) algorithm, achieving a classification accuracy of 88%. This outcome underscores the potential of our 2D light scattering static cytometry for lymphoma cell classification, offering a marker-free, cost-effective approach for early cancer screening at the single-cell level.