A novel automated image analysis system using deep convolutional neural networks can assist to differentiate MDS and AA

Konobu Kimura, Yoko Tabe, Tomohiko Ai, Ikki Takehara, Hiroshi Fukuda, Hiromizu Takahashi, Toshio Naito, Norio Komatsu, Kinya Uchihashi, Akimichi Ohsaka

Research output: Contribution to journalArticle

Abstract

Detection of dysmorphic cells in peripheral blood (PB) smears is essential in diagnostic screening of hematological diseases. Myelodysplastic syndromes (MDS) are hematopoietic neoplasms characterized by dysplastic and ineffective hematopoiesis, which diagnosis is mainly based on morphological findings of PB and bone marrow. We developed an automated diagnostic support system of MDS by combining an automated blood cell image-recognition system using a deep learning system (DLS) powered by convolutional neural networks (CNNs) with a decision-making system using extreme gradient boosting (XGBoost). The DLS of blood cell image-recognition has been trained using datasets consisting of 695,030 blood cell images taken from 3,261 PB smears including hematopoietic malignancies. The DLS simultaneously classified 17 blood cell types and 97 morphological features of such cells with >93.5% sensitivity and >96.0% specificity. The automated MDS diagnostic system successfully differentiated MDS from aplastic anemia (AA) with high accuracy; 96.2% of sensitivity and 100% of specificity (AUC 0.990). This is the first CNN-based automated initial diagnostic system for MDS using PB smears, which is applicable to develop new automated diagnostic systems for various hematological disorders.

Original languageEnglish (US)
Number of pages1
JournalScientific reports
Volume9
Issue number1
DOIs
StatePublished - Sep 16 2019

Fingerprint

Aplastic Anemia
Myelodysplastic Syndromes
Blood Cells
Learning
Hematologic Neoplasms
Hematologic Diseases
Hematopoiesis
Area Under Curve
Decision Making
Bone Marrow
Sensitivity and Specificity
Recognition (Psychology)

ASJC Scopus subject areas

  • General

Cite this

A novel automated image analysis system using deep convolutional neural networks can assist to differentiate MDS and AA. / Kimura, Konobu; Tabe, Yoko; Ai, Tomohiko; Takehara, Ikki; Fukuda, Hiroshi; Takahashi, Hiromizu; Naito, Toshio; Komatsu, Norio; Uchihashi, Kinya; Ohsaka, Akimichi.

In: Scientific reports, Vol. 9, No. 1, 16.09.2019.

Research output: Contribution to journalArticle

Kimura, K, Tabe, Y, Ai, T, Takehara, I, Fukuda, H, Takahashi, H, Naito, T, Komatsu, N, Uchihashi, K & Ohsaka, A 2019, 'A novel automated image analysis system using deep convolutional neural networks can assist to differentiate MDS and AA', Scientific reports, vol. 9, no. 1. https://doi.org/10.1038/s41598-019-49942-z
Kimura, Konobu ; Tabe, Yoko ; Ai, Tomohiko ; Takehara, Ikki ; Fukuda, Hiroshi ; Takahashi, Hiromizu ; Naito, Toshio ; Komatsu, Norio ; Uchihashi, Kinya ; Ohsaka, Akimichi. / A novel automated image analysis system using deep convolutional neural networks can assist to differentiate MDS and AA. In: Scientific reports. 2019 ; Vol. 9, No. 1.
@article{e8215a3c6c584ce6affdc430f4afc923,
title = "A novel automated image analysis system using deep convolutional neural networks can assist to differentiate MDS and AA",
abstract = "Detection of dysmorphic cells in peripheral blood (PB) smears is essential in diagnostic screening of hematological diseases. Myelodysplastic syndromes (MDS) are hematopoietic neoplasms characterized by dysplastic and ineffective hematopoiesis, which diagnosis is mainly based on morphological findings of PB and bone marrow. We developed an automated diagnostic support system of MDS by combining an automated blood cell image-recognition system using a deep learning system (DLS) powered by convolutional neural networks (CNNs) with a decision-making system using extreme gradient boosting (XGBoost). The DLS of blood cell image-recognition has been trained using datasets consisting of 695,030 blood cell images taken from 3,261 PB smears including hematopoietic malignancies. The DLS simultaneously classified 17 blood cell types and 97 morphological features of such cells with >93.5{\%} sensitivity and >96.0{\%} specificity. The automated MDS diagnostic system successfully differentiated MDS from aplastic anemia (AA) with high accuracy; 96.2{\%} of sensitivity and 100{\%} of specificity (AUC 0.990). This is the first CNN-based automated initial diagnostic system for MDS using PB smears, which is applicable to develop new automated diagnostic systems for various hematological disorders.",
author = "Konobu Kimura and Yoko Tabe and Tomohiko Ai and Ikki Takehara and Hiroshi Fukuda and Hiromizu Takahashi and Toshio Naito and Norio Komatsu and Kinya Uchihashi and Akimichi Ohsaka",
year = "2019",
month = "9",
day = "16",
doi = "10.1038/s41598-019-49942-z",
language = "English (US)",
volume = "9",
journal = "Scientific Reports",
issn = "2045-2322",
publisher = "Nature Publishing Group",
number = "1",

}

TY - JOUR

T1 - A novel automated image analysis system using deep convolutional neural networks can assist to differentiate MDS and AA

AU - Kimura, Konobu

AU - Tabe, Yoko

AU - Ai, Tomohiko

AU - Takehara, Ikki

AU - Fukuda, Hiroshi

AU - Takahashi, Hiromizu

AU - Naito, Toshio

AU - Komatsu, Norio

AU - Uchihashi, Kinya

AU - Ohsaka, Akimichi

PY - 2019/9/16

Y1 - 2019/9/16

N2 - Detection of dysmorphic cells in peripheral blood (PB) smears is essential in diagnostic screening of hematological diseases. Myelodysplastic syndromes (MDS) are hematopoietic neoplasms characterized by dysplastic and ineffective hematopoiesis, which diagnosis is mainly based on morphological findings of PB and bone marrow. We developed an automated diagnostic support system of MDS by combining an automated blood cell image-recognition system using a deep learning system (DLS) powered by convolutional neural networks (CNNs) with a decision-making system using extreme gradient boosting (XGBoost). The DLS of blood cell image-recognition has been trained using datasets consisting of 695,030 blood cell images taken from 3,261 PB smears including hematopoietic malignancies. The DLS simultaneously classified 17 blood cell types and 97 morphological features of such cells with >93.5% sensitivity and >96.0% specificity. The automated MDS diagnostic system successfully differentiated MDS from aplastic anemia (AA) with high accuracy; 96.2% of sensitivity and 100% of specificity (AUC 0.990). This is the first CNN-based automated initial diagnostic system for MDS using PB smears, which is applicable to develop new automated diagnostic systems for various hematological disorders.

AB - Detection of dysmorphic cells in peripheral blood (PB) smears is essential in diagnostic screening of hematological diseases. Myelodysplastic syndromes (MDS) are hematopoietic neoplasms characterized by dysplastic and ineffective hematopoiesis, which diagnosis is mainly based on morphological findings of PB and bone marrow. We developed an automated diagnostic support system of MDS by combining an automated blood cell image-recognition system using a deep learning system (DLS) powered by convolutional neural networks (CNNs) with a decision-making system using extreme gradient boosting (XGBoost). The DLS of blood cell image-recognition has been trained using datasets consisting of 695,030 blood cell images taken from 3,261 PB smears including hematopoietic malignancies. The DLS simultaneously classified 17 blood cell types and 97 morphological features of such cells with >93.5% sensitivity and >96.0% specificity. The automated MDS diagnostic system successfully differentiated MDS from aplastic anemia (AA) with high accuracy; 96.2% of sensitivity and 100% of specificity (AUC 0.990). This is the first CNN-based automated initial diagnostic system for MDS using PB smears, which is applicable to develop new automated diagnostic systems for various hematological disorders.

UR - http://www.scopus.com/inward/record.url?scp=85072287371&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85072287371&partnerID=8YFLogxK

U2 - 10.1038/s41598-019-49942-z

DO - 10.1038/s41598-019-49942-z

M3 - Article

C2 - 31527646

AN - SCOPUS:85072287371

VL - 9

JO - Scientific Reports

JF - Scientific Reports

SN - 2045-2322

IS - 1

ER -