Classification of Monocytes, Promonocytes and Monoblasts Using Deep Neural Network Models: An Area of Unmet Need in Diagnostic Hematopathology.

Mazen Osman,Aref Al-Kali,Samia Rizk,Dragan Jevremovic,Zeynettin Akkus,Ahmad Nanaa,Mrinal M Patnaik,Dana Roh,Mohamed E Salama,Phuong L Nguyen

doi:10.3390/jcm10112264

Abstract

The accurate diagnosis of chronic myelomonocytic leukemia (CMML) and acute myeloid leukemia (AML) subtypes with monocytic differentiation relies on the proper identification and quantitation of blast cells and blast-equivalent cells, including promonocytes. This distinction can be quite challenging given the cytomorphologic and immunophenotypic similarities among the monocytic cell precursors. The aim of this study was to assess the performance of convolutional neural networks (CNN) in separating monocytes from their precursors (i.e., promonocytes and monoblasts). We collected digital images of 935 monocytic cells that were blindly reviewed by five experienced morphologists and assigned into three subtypes: monocyte, promonocyte, and blast. The consensus between reviewers was considered as a ground truth reference label for each cell. In order to assess the performance of CNN models, we divided our data into training (70%), validation (10%), and test (20%) datasets, as well as applied fivefold cross validation. The CNN models did not perform well for predicting three monocytic subtypes, but their performance was significantly improved for two subtypes (monocyte vs. promonocytes + blasts). Our findings (1) support the concept that morphologic distinction between monocytic cells of various differentiation level is difficult; (2) suggest that combining blasts and promonocytes into a single category is desirable for improved accuracy; and (3) show that CNN models can reach accuracy comparable to human reviewers (0.78 ± 0.10 vs. 0.86 ± 0.05). As far as we know, this is the first study to separate monocytes from their precursors using CNN.

Highlights

To meet the World Health Organization (WHO) diagnostic criteria, the peripheral blood (PB) or bone marrow (BM) of patients with acute monoblastic and monocytic leukemia must have ≥20% blasts, and ≥80% of the leukemic cells must be of monocytic lineage, including monoblasts, promonocytes, and monocytes
The performance of the five convolutional neural networks (CNN) models with different configurations and the resulting classification of the monocytic cells on the validation and test datasets are shown in Tables 1 and 2
Using configuration 2, the accuracy of CNN models for predicting three subcategories (Table 1) on the test dataset ranged from 42% to 58%, while it ranged from 70% to 85% for predicting two subcategories (Table 2)

Summary

Introduction

The classification of the monocytic subpopulations (monoblasts, promonocytes, and monocytes) is important for the proper diagnosis and classification of various monocyticlineage leukemias, namely, chronic myelomonocytic leukemia (CMML) and acute myeloid leukemia (AML), including acute monoblastic leukemia and acute monocytic leukemia, and acute myelomonocytic leukemia [1]. To meet the World Health Organization (WHO) diagnostic criteria, the peripheral blood (PB) or bone marrow (BM) of patients with acute monoblastic and monocytic leukemia must have ≥20% blasts (including promonocytes), and ≥80% of the leukemic cells must be of monocytic lineage, including monoblasts, promonocytes, and monocytes. Differentiation between acute monoblastic leukemia and acute monocytic leukemia is based on the relative proportions of monoblasts and promonocytes. The majority of the monocytic cells (≥80%) are monoblasts, whereas in acute monocytic leukemia, the predominant populations are mature monocytes and promonocytes [1,2,3]

Objectives

Methods

Results

Discussion

Conclusion