Related Experiment Video
Updated: Dec 18, 2025
12:05
Database-guided Flow-cytometry for Evaluation of Bone Marrow Myeloid Cell Maturation
Published on: November 3, 2018
12.1K
Hematologist-Level Classification of Mature B-Cell Neoplasm Using Deep Learning on Multiparameter Flow Cytometry Data
Max Zhao1,2, Nanditha Mallesh1, Alexander Höllein3,4
1Institute for Genomic Statistics and Bioinformatics, University Bonn, Bonn, Germany.
Summary
Artificial intelligence analyzes multiparameter flow cytometry (MFC) data as images to detect B-cell neoplasms. This AI model accurately distinguishes healthy from diseased samples and classifies seven subtypes with high confidence.
Area of Science:
- Hematology
- Computational Biology
- Artificial Intelligence
Background:
- Multiparameter flow cytometry (MFC) generates extensive data.
- Analyzing MFC data for hematologic malignancies is complex.
- Computer vision offers novel analytical approaches for MFC data.
Purpose of the Study:
- To develop an artificial intelligence (AI) model for analyzing MFC data.
- To classify mature B-cell neoplasms using AI-driven image analysis.
- To assess the diagnostic performance and confidence of the AI model.
Main Methods:
- Transformed MFC raw data into multicolor 2D images using self-organizing maps.
- Classified these images using a convolutional neural network (CNN).
- Trained the AI model on 18,274 cases and validated on 2,348 cases.
Main Results:
- The AI model achieved a weighted F1 score of 0.94 in classifying B-cell neoplasms.
- Successfully differentiated healthy samples from diseased ones.
- Classified 70% of cases with high confidence (≥0.95).
Conclusions:
- AI-based image analysis of MFC data is effective for diagnosing B-cell neoplasms.
- The model demonstrates high accuracy and confidence in classification.
- Further improvements are anticipated with larger datasets, especially for rare subtypes.
