Ekin Tiu

A self-supervised model trained on chest X-ray images that lack explicit annotations performs pathology-classification tasks with accuracies comparable to those of radiologists. nature.com/articles/s4155…

Self-supervised (no expensive detailed annotation) classification of chest X rays pathology "look Ma, no annotations" @natBME go.nature.com/3qDBQyd in a collaboration across @Stanford @HarvardDBMI w @pranavrajpurkar @curtlanglotz @AndrewYNg and Ekin Tiu

Great coverage of our recent #CheXzero publication for self-supervised learning using radiology images and reports. Coverage by @EkaterinaPeshev @harvardmed hms.harvard.edu/news/no-labels…

Today, we're publishing CheXzero in @natBME 🚀 We show we can detect diseases from chest X-ray images at the level of radiologists *without labeling for individual diseases*. How do we do it? By teaching the model to read the radiology report. nature.com/articles/s4155…