Abstract
Acute lymphoblastic leukemia (ALL) is the most common childhood cancer. While there are a number of well‐recognized prognostic biomarkers at diagnosis, the most powerful independent prognostic factor is the response of the leukemia to induction chemotherapy (Campana and Pui: Blood 129 (2017) 1913–1918). Given the potential for machine learning to improve precision medicine, we tested its capacity to monitor disease in children undergoing ALL treatment. Diagnostic and on‐treatment bone marrow samples were labeled with an ALL‐discriminating antibody combination and analyzed by imaging flow cytometry. Ignoring the fluorescent markers and using only features extracted from bright‐field and dark‐field cell images, a deep learning model was able to identify ALL cells at an accuracy of >88%. This antibody‐free, single cell method is cheap, quick, and could be adapted to a simple, laser‐free cytometer to allow automated, point‐of‐care testing to detect slow early responders. Adaptation to other types of leukemia is feasible, which would revolutionize residual disease monitoring. © 2020 The Authors. Cytometry Part A published by Wiley Periodicals, Inc. on behalf of International Society for Advancement of Cytometry.
Highlights
Flow cytometry has been integrated with fluorescence microscopy to create imaging flow cytometry (IFC), where an image of each cell is captured as it flows past a light source and a charge-coupled device (CCD) detector [9]
The highcontent data rapidly captured using IFC is well-suited to classification of cell phenotypes by machine learning, deep learning, given the large number of training images required to apply deep convolutional neural networks
Using diagnostic and follow-up bone marrow aspirates taken during remission induction from children with B lineage Acute lymphoblastic leukemia (ALL) and with the leukemia-associated immunophenotype (n = 30, collected over two and a half years), we first trained a convolutional neural network to separate IFC cell images into three classes: ALL blasts (CD19+CD10+CD34+/− and CD45+/−), normal B lymphocytes (CD19+CD10−CD34− and CD45+), and “other” nucleated cells—denoting granulocytes, monocytes, deformed/dead cells etc.)
Summary
This is a four to ten-antibody assay, in which the leukemic cells fall into so-called empty spaces within scatter plots, distinct from regions housing normal lymphocyte progenitors (Fig. S1 and S2). It can discriminate and quantify leukemic cells in “on-treatment” samples [7]. Both methods of ALL detection are highly specialized, require specific reagents and extensive training, and are slow, labor intensive, and costly. We wondered whether deep convolutional neural networks might be able to detect leukemic cells from bone marrow samples of ALL patients using few or no fluorescent markers
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
