TY - JOUR
T1 - A deep-learning algorithm to classify skin lesions from mpox virus infection
AU - Thieme, Alexander H.
AU - Zheng, Yuanning
AU - Machiraju, Gautam
AU - Sadee, Chris
AU - Mittermaier, Mirja
AU - Gertler, Maximilian
AU - Salinas, Jorge L.
AU - Srinivasan, Krithika
AU - Gyawali, Prashnna
AU - Carrillo-Perez, Francisco
AU - Capodici, Angelo
AU - Uhlig, Maximilian
AU - Habenicht, Daniel
AU - Löser, Anastassia
AU - Kohler, Maja
AU - Schuessler, Maximilian
AU - Kaul, David
AU - Gollrad, Johannes
AU - Ma, Jackie
AU - Lippert, Christoph
AU - Billick, Kendall
AU - Bogoch, Isaac
AU - Hernandez-Boussard, Tina
AU - Geldsetzer, Pascal
AU - Gevaert, Olivier
N1 - Funding Information:
We are very grateful for the support of N. Attkinson (DermNet NZ) for providing high quality non-MPXV images used for training the MPXV-CNN. We thank very much N. Veien (Danderm) and C.D. Verros (Hellenic Dermatological Atlas) for providing their great dermatological repositories and their active support for this project. We thank J. Benzler for his valuable suggestions for this manuscript and project. We thank the open-source community for their contributions to PoxApp. We thank I. Giret for her contributions to the table of this manuscript. G.M. is grateful for institutional support from Stanford Data Science and Biomedical Informatics Training Program at Stanford 2T15LM007033. F.C.P. was supported by the Spanish Ministry of Sciences, Innovation, and Universities under Projects RTI-2018-101674-B-I00 and PID2021-128317OB-I00, the project from J.de Andalucia P20-00163 and a Predoctoral scholarship from the Fulbright Spanish Commission. M.S. was supported by the ERP scholarship funded by the German Federal Ministry for Economic Affairs and Climate Action and Studienstiftung des deutschen Volkes (German Academic Scholarship Foundation). P.G. is a Chan Zuckerberg Biohub investigator and was supported by NIH grant DP2AI171011. J.L.S. was supported by NIH grant 5R25AI147369-03. A.H.T., C.L. and J.M. were supported by the German Federal Ministry for Economic Affairs and Climate Action (BMWi) under the project DAKI-FWS (BMWi 01MK21009E). A.H.T. and M.M. are both participants in the BIH—Charité Digital Clinician Scientist Program funded by the Charité—Universitätsmedizin Berlin, the Berlin Institute of Health and the German Research Foundation (DFG).
Funding Information:
This project has been supported by funding from the German Federal Ministry for Economic Affairs and Climate Action (BMWi) under the project DAKI-FWS (BMWi 01MK21009E).
Publisher Copyright:
© 2023, The Author(s).
PY - 2023/3
Y1 - 2023/3
N2 - Undetected infection and delayed isolation of infected individuals are key factors driving the monkeypox virus (now termed mpox virus or MPXV) outbreak. To enable earlier detection of MPXV infection, we developed an image-based deep convolutional neural network (named MPXV-CNN) for the identification of the characteristic skin lesions caused by MPXV. We assembled a dataset of 139,198 skin lesion images, split into training/validation and testing cohorts, comprising non-MPXV images (n = 138,522) from eight dermatological repositories and MPXV images (n = 676) from the scientific literature, news articles, social media and a prospective cohort of the Stanford University Medical Center (n = 63 images from 12 patients, all male). In the validation and testing cohorts, the sensitivity of the MPXV-CNN was 0.83 and 0.91, the specificity was 0.965 and 0.898 and the area under the curve was 0.967 and 0.966, respectively. In the prospective cohort, the sensitivity was 0.89. The classification performance of the MPXV-CNN was robust across various skin tones and body regions. To facilitate the usage of the algorithm, we developed a web-based app by which the MPXV-CNN can be accessed for patient guidance. The capability of the MPXV-CNN for identifying MPXV lesions has the potential to aid in MPXV outbreak mitigation.
AB - Undetected infection and delayed isolation of infected individuals are key factors driving the monkeypox virus (now termed mpox virus or MPXV) outbreak. To enable earlier detection of MPXV infection, we developed an image-based deep convolutional neural network (named MPXV-CNN) for the identification of the characteristic skin lesions caused by MPXV. We assembled a dataset of 139,198 skin lesion images, split into training/validation and testing cohorts, comprising non-MPXV images (n = 138,522) from eight dermatological repositories and MPXV images (n = 676) from the scientific literature, news articles, social media and a prospective cohort of the Stanford University Medical Center (n = 63 images from 12 patients, all male). In the validation and testing cohorts, the sensitivity of the MPXV-CNN was 0.83 and 0.91, the specificity was 0.965 and 0.898 and the area under the curve was 0.967 and 0.966, respectively. In the prospective cohort, the sensitivity was 0.89. The classification performance of the MPXV-CNN was robust across various skin tones and body regions. To facilitate the usage of the algorithm, we developed a web-based app by which the MPXV-CNN can be accessed for patient guidance. The capability of the MPXV-CNN for identifying MPXV lesions has the potential to aid in MPXV outbreak mitigation.
UR - http://www.scopus.com/inward/record.url?scp=85149145557&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85149145557&partnerID=8YFLogxK
U2 - 10.1038/s41591-023-02225-7
DO - 10.1038/s41591-023-02225-7
M3 - Article
C2 - 36864252
AN - SCOPUS:85149145557
SN - 1078-8956
VL - 29
SP - 738
EP - 747
JO - Nature medicine
JF - Nature medicine
IS - 3
ER -