Case Study: King's College London achieves 6x faster annotation with Encord

Overcoming The Mental Torture of Manual Data Labelling

King's College London, in collaboration with a gastroenterology director at King's College Hospital, faced the immense challenge of manual data labeling for medical AI research. The process of annotating colonoscopy videos to train computer vision models for polyp detection was described as "mental torture," being prohibitively time-consuming and an inefficient use of highly trained clinicians' expertise. They partnered with Encord and utilized its platform for data-centric computer vision to address this.

Encord's platform provided intuitive, built-in annotation tools and micromodels that automated the labeling process. This solution enabled the medical team to annotate data over six times faster than with traditional methods, with the platform automatically generating 97% of the labels. The clinicians' role shifted from manual annotators to quality assurance reviewers, dramatically increasing efficiency and allowing their expertise to be used more effectively while significantly accelerating their AI research.

King's College London

Bu Hayee

Director for Gastroenterology and Endoscopy