TY - JOUR
T1 - Interpretable multimodal deep learning for real-time pan-tissue pan-disease pathology search on social media
AU - Schaumberg, Andrew J.
AU - Juarez-Nicanor, Wendy C.
AU - Choudhury, Sarah J.
AU - Pastrián, Laura G.
AU - Pritt, Bobbi S.
AU - Prieto Pozuelo, Mario
AU - Sotillo Sánchez, Ricardo
AU - Ho, Khanh
AU - Zahra, Nusrat
AU - Sener, Betul Duygu
AU - Yip, Stephen
AU - Xu, Bin
AU - Annavarapu, Srinivas Rao
AU - Morini, Aurélien
AU - Jones, Karra A.
AU - Rosado-Orozco, Kathia
AU - Mukhopadhyay, Sanjay
AU - Miguel, Carlos
AU - Yang, Hongyu
AU - Rosen, Yale
AU - Ali, Rola H.
AU - Folaranmi, Olaleke O.
AU - Gardner, Jerad M.
AU - Rusu, Corina
AU - Stayerman, Celina
AU - Gross, John
AU - Suleiman, Dauda E.
AU - Sirintrapun, S. Joseph
AU - Aly, Mariam
AU - Fuchs, Thomas J.
N1 - Funding Information:
Acknowledgements AJS thanks Dr. Marcus Lambert and Pedro Cito Silberman for organizing the Weill Cornell High School Science Immersion Program. AJS thanks Terrie Wheeler and the Weill Cornell Medicine Samuel J. Wood Library for providing vital space for AJS, WCJ, and SJC to work early in this project. AJS thanks Dr. Joanna Cyrta of Institut Curie for H&E-saffron (HES) discussion. AJS thanks Dr. Takehiko Fujisawa of Chiba University for his free pathology photos contributed to social media and this project via @Patholwalker on Twitter. AJS was supported by NIH/NCI grant F31CA214029 and the Tri-Institutional Training Program in Computational Biology and Medicine (via NIH training grant T32GM083937). This research was funded in part through the NIH/NCI Cancer Center Support Grant P30CA008748. We are grateful to the patients who made this study possible.
Publisher Copyright:
© 2020, The Author(s).
PY - 2020/11/1
Y1 - 2020/11/1
N2 - Pathologists are responsible for rapidly providing a diagnosis on critical health issues. Challenging cases benefit from additional opinions of pathologist colleagues. In addition to on-site colleagues, there is an active worldwide community of pathologists on social media for complementary opinions. Such access to pathologists worldwide has the capacity to improve diagnostic accuracy and generate broader consensus on next steps in patient care. From Twitter we curate 13,626 images from 6,351 tweets from 25 pathologists from 13 countries. We supplement the Twitter data with 113,161 images from 1,074,484 PubMed articles. We develop machine learning and deep learning models to (i) accurately identify histopathology stains, (ii) discriminate between tissues, and (iii) differentiate disease states. Area Under Receiver Operating Characteristic (AUROC) is 0.805–0.996 for these tasks. We repurpose the disease classifier to search for similar disease states given an image and clinical covariates. We report precision@k = 1 = 0.7618 ± 0.0018 (chance 0.397 ± 0.004, mean ±stdev). The classifiers find that texture and tissue are important clinico-visual features of disease. Deep features trained only on natural images (e.g., cats and dogs) substantially improved search performance, while pathology-specific deep features and cell nuclei features further improved search to a lesser extent. We implement a social media bot (@pathobot on Twitter) to use the trained classifiers to aid pathologists in obtaining real-time feedback on challenging cases. If a social media post containing pathology text and images mentions the bot, the bot generates quantitative predictions of disease state (normal/artifact/infection/injury/nontumor, preneoplastic/benign/low-grade-malignant-potential, or malignant) and lists similar cases across social media and PubMed. Our project has become a globally distributed expert system that facilitates pathological diagnosis and brings expertise to underserved regions or hospitals with less expertise in a particular disease. This is the first pan-tissue pan-disease (i.e., from infection to malignancy) method for prediction and search on social media, and the first pathology study prospectively tested in public on social media. We will share data through http://pathobotology.org. We expect our project to cultivate a more connected world of physicians and improve patient care worldwide.
AB - Pathologists are responsible for rapidly providing a diagnosis on critical health issues. Challenging cases benefit from additional opinions of pathologist colleagues. In addition to on-site colleagues, there is an active worldwide community of pathologists on social media for complementary opinions. Such access to pathologists worldwide has the capacity to improve diagnostic accuracy and generate broader consensus on next steps in patient care. From Twitter we curate 13,626 images from 6,351 tweets from 25 pathologists from 13 countries. We supplement the Twitter data with 113,161 images from 1,074,484 PubMed articles. We develop machine learning and deep learning models to (i) accurately identify histopathology stains, (ii) discriminate between tissues, and (iii) differentiate disease states. Area Under Receiver Operating Characteristic (AUROC) is 0.805–0.996 for these tasks. We repurpose the disease classifier to search for similar disease states given an image and clinical covariates. We report precision@k = 1 = 0.7618 ± 0.0018 (chance 0.397 ± 0.004, mean ±stdev). The classifiers find that texture and tissue are important clinico-visual features of disease. Deep features trained only on natural images (e.g., cats and dogs) substantially improved search performance, while pathology-specific deep features and cell nuclei features further improved search to a lesser extent. We implement a social media bot (@pathobot on Twitter) to use the trained classifiers to aid pathologists in obtaining real-time feedback on challenging cases. If a social media post containing pathology text and images mentions the bot, the bot generates quantitative predictions of disease state (normal/artifact/infection/injury/nontumor, preneoplastic/benign/low-grade-malignant-potential, or malignant) and lists similar cases across social media and PubMed. Our project has become a globally distributed expert system that facilitates pathological diagnosis and brings expertise to underserved regions or hospitals with less expertise in a particular disease. This is the first pan-tissue pan-disease (i.e., from infection to malignancy) method for prediction and search on social media, and the first pathology study prospectively tested in public on social media. We will share data through http://pathobotology.org. We expect our project to cultivate a more connected world of physicians and improve patient care worldwide.
UR - http://www.scopus.com/inward/record.url?scp=85085513830&partnerID=8YFLogxK
U2 - 10.1038/s41379-020-0540-1
DO - 10.1038/s41379-020-0540-1
M3 - Article
C2 - 32467650
AN - SCOPUS:85085513830
SN - 0893-3952
VL - 33
SP - 2169
EP - 2185
JO - Modern Pathology
JF - Modern Pathology
IS - 11
ER -