Abstract
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 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 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, pre-neoplastic/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 pathobotology.org. We expect our project to cultivate a more connected world of physicians and improve patient care worldwide.
Footnotes
↵β These pathologist authors generously donated cases.
↵δ These authors are Principal Investigators of this work.
↵a ajs625{at}cornell.edu (Twitter @schaumberg_a)
↵b ma3631{at}columbia.edu (Twitter @mariam_s_aly)
↵c fuchst{at}mskcc.org (Twitter @ThomasFuchsAI).
Link to site to host data http://pathobotology.org Additional cluster and importance analyses. Improved performance.
↵1 ImageIO documentation available here: https://docs.oracle.com/javase/7/docs/api/javax/imageio/Image10.html
↵2 Courts in the United States have ruled that images posted to social media are still owned by their authors and are not public domain. Indeed, in Morel v. AFP, AFP was ordered to pay Morel $1.2 million for copyright infringement because AFP used images that Morel posted to social media.
↵3 Normal cerebellum case by S.Y. at https://twitter.com/Sty_md/status/821840894634565632
↵4 A case of this is from author K.H., where a different pathologist gave the diagnosis, and he agreed. We summarized this as “metastatic lobular carcinoma” in the auxiliary annotation file for the tweet https://twitter.com/Ho_Khanh_MD/status/999989201734197250.
↵5 A case of this is from author M.P.P., where M.P.P. wrote “IDC DIN LISN” directly on a shared histology image in the tweet https://twitter.com/dr_MPrieto/status/890118713155997696 so we wrote this text in the auxiliary annotation file for the tweet.
↵6 A case of this is from K.H., observing iron pill lesions in stomach biopsy https://twitter.com/Ho_Khanh_MD/status/963800933716123648.
↵7 For this formula please see https://github.com/keras-team/keras/issues/6444
↵8 Case at https://twitter.com/BinXu16/status/980404471833313280 “Kudo to @drkennethtang @luishcruzc and @DrGeeONE The answer of this case can be seen in the right corner of the 3rd picture. Dx: Echinococcus (hydatid cyst) with necrotizing pneumonia, abscess, and granulomatous inflammation. Additional high power pictures attached.”