Opportunity
For patients with eye disease such as macular degeneration, the macula needs to be scanned and analysed at every appointment. Macuject’s software combines AI and real-world expertise to help clinicians analyse the scans and take the appropriate action. The Macuject team needed a machine learning model to underpin the software, and an easy-to-understand labelling platform for experts to annotate data remotely. 

Solution
Silverpond used HighLighter to train machine learning models to detect signs of macular degeneration. A custom plugin for HighLighter’s labelling tool was created based on the relationship between layers in the 3D scans. HighLighter Quality Assurance tools adjusted for subjectivity in doctors’ labels, resulting in a more accurate model

Results
-The model detects three types of pathology that are indicators of disease progression
-Potential to improve patient outcomes and clinician efficiency – the model is currently being tested by clinicians in practice 
-HighLighter’s flexible labelling tool supported remote labelling, improved efficiency, and provided custom industry-specific support
-Resulted in two patents pending for Macuject