npj Digital Medicine (Jan 2021)
Deep COVID DeteCT: an international experience on COVID-19 lung detection and prognosis using chest CT
- Edward H. Lee,
- Jimmy Zheng,
- Errol Colak,
- Maryam Mohammadzadeh,
- Golnaz Houshmand,
- Nicholas Bevins,
- Felipe Kitamura,
- Emre Altinmakas,
- Eduardo Pontes Reis,
- Jae-Kwang Kim,
- Chad Klochko,
- Michelle Han,
- Sadegh Moradian,
- Ali Mohammadzadeh,
- Hashem Sharifian,
- Hassan Hashemi,
- Kavous Firouznia,
- Hossien Ghanaati,
- Masoumeh Gity,
- Hakan Doğan,
- Hojjat Salehinejad,
- Henrique Alves,
- Jayne Seekins,
- Nitamar Abdala,
- Çetin Atasoy,
- Hamidreza Pouraliakbar,
- Majid Maleki,
- S. Simon Wong,
- Kristen W. Yeom
Affiliations
- Edward H. Lee
- Department of Radiology, School of Medicine, Stanford University
- Jimmy Zheng
- Department of Radiology, School of Medicine, Stanford University
- Errol Colak
- Unity Health Toronto, University of Toronto
- Maryam Mohammadzadeh
- Division of Radiology, Amir Alam Hospital, Tehran University of Medical Sciences
- Golnaz Houshmand
- Rajaie Cardiovascular Medical and Research Center, Iran University of Medical Sciences
- Nicholas Bevins
- Henry Ford Health System
- Felipe Kitamura
- Universidade Federal de São Paulo (UNIFESP)
- Emre Altinmakas
- Department of Radiology, Koç University School of Medicine
- Eduardo Pontes Reis
- Hospital Israelita Albert Einstein
- Jae-Kwang Kim
- Department of Radiology, School of Medicine, Kyungpook National University
- Chad Klochko
- Rajaie Cardiovascular Medical and Research Center, Iran University of Medical Sciences
- Michelle Han
- Department of Radiology, School of Medicine, Stanford University
- Sadegh Moradian
- School of Medicine, Tehran University of Medical Sciences
- Ali Mohammadzadeh
- Rajaie Cardiovascular Medical and Research Center, Iran University of Medical Sciences
- Hashem Sharifian
- Division of Radiology, Amir Alam Hospital, Tehran University of Medical Sciences
- Hassan Hashemi
- Advanced Diagnostic and Interventional Radiology Research Center(ADIR), Medical Imaging Center, Imam Khomeini Hospital Complex, Tehran University of Medical Sciences
- Kavous Firouznia
- Advanced Diagnostic and Interventional Radiology Research Center(ADIR), Medical Imaging Center, Imam Khomeini Hospital Complex, Tehran University of Medical Sciences
- Hossien Ghanaati
- Advanced Diagnostic and Interventional Radiology Research Center(ADIR), Medical Imaging Center, Imam Khomeini Hospital Complex, Tehran University of Medical Sciences
- Masoumeh Gity
- Advanced Diagnostic and Interventional Radiology Research Center(ADIR), Medical Imaging Center, Imam Khomeini Hospital Complex, Tehran University of Medical Sciences
- Hakan Doğan
- Department of Radiology, Koç University School of Medicine
- Hojjat Salehinejad
- Unity Health Toronto, University of Toronto
- Henrique Alves
- Universidade Federal de São Paulo (UNIFESP)
- Jayne Seekins
- Department of Radiology, School of Medicine, Stanford University
- Nitamar Abdala
- Universidade Federal de São Paulo (UNIFESP)
- Çetin Atasoy
- Department of Radiology, Koç University School of Medicine
- Hamidreza Pouraliakbar
- Rajaie Cardiovascular Medical and Research Center, Iran University of Medical Sciences
- Majid Maleki
- Rajaie Cardiovascular Medical and Research Center, Iran University of Medical Sciences
- S. Simon Wong
- Department of Electrical Engineering, Stanford University
- Kristen W. Yeom
- Department of Radiology, School of Medicine, Stanford University
- DOI
- https://doi.org/10.1038/s41746-020-00369-1
- Journal volume & issue
-
Vol. 4,
no. 1
pp. 1 – 11
Abstract
Abstract The Coronavirus disease 2019 (COVID-19) presents open questions in how we clinically diagnose and assess disease course. Recently, chest computed tomography (CT) has shown utility for COVID-19 diagnosis. In this study, we developed Deep COVID DeteCT (DCD), a deep learning convolutional neural network (CNN) that uses the entire chest CT volume to automatically predict COVID-19 (COVID+) from non-COVID-19 (COVID−) pneumonia and normal controls. We discuss training strategies and differences in performance across 13 international institutions and 8 countries. The inclusion of non-China sites in training significantly improved classification performance with area under the curve (AUCs) and accuracies above 0.8 on most test sites. Furthermore, using available follow-up scans, we investigate methods to track patient disease course and predict prognosis.
