School of Engineering and Computer Science

Automatic leukocoria detection

This page is about our research on automatic detection of leukocoria (white-eye reflection) using software. Our research is producing algorithms which can learn from examples to automatically detect when a photo contains a white-eye reflection that may be an indication of a solid-tumor pediatric cancer called Retinoblastoma. Here is an example of an eye that appears leukocoric:

Our research involves analyzing images of eyes detected within larger natural images. As such, we are especially interested in separating true leukocoria (caused by diseases such as Retinoblastoma) from normal (black and red) pupils, as well as pseudoleukocoria (caused by some camera flashes, but not indicating a negative condition).

iPad / iPhone app

You can download our white-eye detector app from the Apple app store! Please give it a try.

We previously had a web-based version of this software available, but the app version has much better software support. Please use that version.




Dr. Bryan Shaw on TEDxUtica (November 7, 2014)