OCT-AID-lite brings automated plaque segmentation closer to real-time clinical use

A new study led by Ruben van der Waerden introduces OCT-AID-lite, a lightweight deep-learning model for near real-time multi-class segmentation of intravascular optical coherence tomography (OCT) images.

Using knowledge distillation and semi-supervised learning, OCT-AID-lite substantially reduces processing time while maintaining accurate segmentation of coronary plaque and vessel structures. The approach brings automated quantitative OCT analysis closer to real-time clinical use.

The work was published in the European Heart Journal – Digital Health.

👉 Read the full paper here!