Discovery of Phenotypic Properties of PBMCs from Healthy Donors & AML Patients Using High-Content Phenotypic Imaging and AI-guided Analysis

Abstract

Acute myeloid leukemia (AML) is one of the heterogeneous hematological malignancies that originate in the bone marrow, causing abnormally differentiated precursor cells of myeloid lineage to proliferate in the blood system. The condition is associated with severe complications in the patients, resulting in approximately 130 000 deaths globally. The purpose of this MSc thesis was to investigate the phenotypic properties of peripheral blood mononuclear cells (PBMCs) isolated from peripheral blood of the healthy donors and AML patients by utilising high-content (HC) single-cell imaging and deep learning (DL) -based image analysis methods. This project particularly aimed to answer the question of whether the PBMCs isolated from AML patients could be differentiated from the healthy PBMCs using the HC phenotypic imaging assay (Cell Painting) and the image analysis pipeline, which was initially established for healthy PBMCs. The HC phenotypic imaging pipeline is composed of sample handling and preparation of cell delivery to the imaging plates, fluorescence staining of cells with the Cell Painting assay, imaging with HC fluorescence confocal microscope (FIMM High Content Imaging and Analysis unit, HCA), followed by DL-based image analysis and phenotype data visualization using UMAP and PCA methods. The results indicated that the pipeline was able to differentiate PBMCs of AML patients from healthy PBMCs in the UMAP embedding based on the captured and extracted cell phenotypes. The pipeline showed capabilities to generalize on cells derived from leukemia patients, which could potentially give new insights into the nature of acute leukemias and the classification of their subtypes in the future.