Format

Send to

Choose Destination
J Pathol Inform. 2018 Apr 20;9:16. doi: 10.4103/jpi.jpi_76_17. eCollection 2018.

A Method for the Interpretation of Flow Cytometry Data Using Genetic Algorithms.

Author information

1
Logical Cytometry, Atlanta GA, USA.

Abstract

Background:

Flow cytometry analysis is the method of choice for the differential diagnosis of hematologic disorders. It is typically performed by a trained hematopathologist through visual examination of bidimensional plots, making the analysis time-consuming and sometimes too subjective. Here, a pilot study applying genetic algorithms to flow cytometry data from normal and acute myeloid leukemia subjects is described.

Subjects and Methods:

Initially, Flow Cytometry Standard files from 316 normal and 43 acute myeloid leukemia subjects were transformed into multidimensional FITS image metafiles. Training was performed through introduction of FITS metafiles from 4 normal and 4 acute myeloid leukemia in the artificial intelligence system.

Results:

Two mathematical algorithms termed 018330 and 025886 were generated. When tested against a cohort of 312 normal and 39 acute myeloid leukemia subjects, both algorithms combined showed high discriminatory power with a receiver operating characteristic (ROC) curve of 0.912.

Conclusions:

The present results suggest that machine learning systems hold a great promise in the interpretation of hematological flow cytometry data.

KEYWORDS:

Flow cytometry; image analysis; leukemia; machine learning

Supplemental Content

Full text links

Icon for Medknow Publications and Media Pvt Ltd Icon for PubMed Central
Loading ...
Support Center