Research

Efficient OCT volume reconstruction from slit lamp microscopes

We developed a novel strategy to build a low-cost slit-lamp integrated OCT device that allows for 3D OCT imaging.

Slit-lamp based video mosaicking

Slitlamp microscopes allow fast inspection of the eye retina, but small fields of view. We explore how large panoramas of the retina can be constructed in this challenging application.

Vision-based surgical instrument detection and tracking

In the context of robotic and imaging control, automatic localization of surgical instruments in a field of view is critical for optimal surgical procedures. We have developed a number of new strategies to do so effectively in challengning imaging conditions. 

OCT-based SRT dosimetry control

SRT laser treatment is invisible to the naked eye but perceived via OCT imaging. We look for novel dosimetry strategies for this promising treatment laser. 

Active Learning in Biomedical imaging

Active selection of annotations is time consuming and error prone. We explore how adaptive and active algorithmic strategies can reduce the annotation burden for experts. 

Automatic Eye and Retinoblastoma segmentation in MRI

Automatic segmentation of eye structures, both healthy and pathological, are critical for both treatment planning and documentation. We are thus interested in effective machine learning strategies to do so effectively.