Background 

Leukemia is a common cancer found in blood cells and bone marrow. A timely diagnosis greatly increases an individual’s prognosis and likelihood of survival. Single-cell flow cytometry (FCM) is used for diagnosis due to its capability to measure the size, shape, and intracellular markers within single cells. A pathologist or technician categorizes the cells as cancerous or non-cancerous using a process called manual gating, which depends on slowly filtering out cells based on expected cell distributions for different intracellular markers.

Automating flow cytometry data interpretation with machine learning can provide rapid, accurate, and cost-effective leukemia diagnosis, by mitigating human error and enhancing accuracy

Algorithms automate the process of standard manual gating for higher throughput data collection. DAFI and FlowSOM are automated gating methods that typically use traditional bioinformatics algorithms like k-means clustering and k-nearest neighbors. Subtypes of leukemia can be easily diagnosed through the use of machine learning techniques as well. This allows medical professionals to determine the subtype specific mortality risk

Credit: Kiran Kumar