AI to detect papilledema from ocular fundus photographs

April 2020 – Papilledema is optic disc swelling due to high intracranial pressure possibly due to many factors such as cerebral hemorrhage, head trauma, meningitis, hydrocephalus, spinal cord lesions, impairment of cerebral sinus drainage and idiopathic intracranial hypertension (IIH). Irrespective of the cause, visual loss is the feared morbidity of papilledema, and the main mechanism of optic nerve damage is intraneuronal ischemia secondary to axoplasmic flow stasis.

Recently, SERI researchers conducted a study where they trained, validated, and externally tested a deep-learning system (DLS) to classify optic disks as being normal or having papilledema or other abnormalities. The size/scope of the study covered:

  • Nearly 16,000 retrospectively collected ocular fundus photographs obtained with pharmacologic pupillary dilation
  • Across multiple ethnic populations from 11 countries
  • Using various digital cameras

In the validation set, the system discriminated

  • optic disks with papilledema from normal disks and disks with non-papilledema abnormalities with an AUC of 0.99 (95% confidence interval [CI], 0.98 to 0.99)
  • normal from abnormal disks with an AUC of 0.99 (95% CI, 0.99 to 0.99)

A deep-learning system using fundus photographs with pharmacologically dilated pupils differentiated among optic disks with papilledema, normal disks, and disks with nonpapilledema abnormalities.


This article was originally published on New England Journal of Medicine on 30 April 2020 and can be read at:


About New England Journal of Medicine

The New England Journal of Medicine (NEJM) is the world’s leading medical journal and website. Published continuously for over 200 years, NEJM delivers high-quality, peer-reviewed research and interactive clinical content to physicians, educators, and the global medical community.

Their mission is to bring physicians the best research and information at the intersection of biomedical science and clinical practice and to present this information in understandable and clinically useful formats that inform health care delivery and improve patient outcomes.



Please direct all queries related to this article to Marketing at


Comments are closed.