Science blog

Improving cytologic screening using AI-powered workflow

01 Jan 2020

The application of Convolutional Neural Networks (CNN), an advanced form of Artificial Intelligence (AI), has been widely used in research for automating cytology images. Most of the research to date however, focuses on training the Neutral Networks to classify individual cells or regions of the image.

Recent advances in the CNN development is showing the potential in using AI to deliver a diagnosis for the entire slide in routine laboratory tests.

With this, AI has truly become both a productive tool and a diagnostic aid for the cytopathologic screening for cancer and other diseases. In this AI-powered workflow for screening tests, all slides are first screened by the computer, where more than 90% of the cases are likely to be normal. The computer produces a diagnostic score for each case, where the cytotechnician would then select only the cases that are most likely to be abnormal to review further, thus enabling them to focus on the patients with the most critical need. Cases classified as normal would be subjected to QC manual review.


Another advancement in the CNN development is the ability to look inside what has been commonly known as the black box. While CNN is trained using whole side images and their associated outcome, data scientists at Keen Eye have devised a way to enable cytologists to look inside the black box of CNN.  For abnormal cases, specific areas on the slide where AI finds the presence of abnormal cells are highlighted, thus drawing the viewer’s attention for fast and decisive final diagnosis.


Sounds too good to be true? This new frontier in cytology tests is no longer just a pipedream.  Join Sylvain Berlemont, Keen Eye CEO, and other industry experts for a panel discussion at the American Society of Cytopathology this Friday, November 15, 8 am to learn how Artificial Intelligence can help improve patient care as a laboratory test aid.