Automatic image quality assurance and root cause prediction of scanning systems
Quality control is of particular importance to a medical imaging company such as Optos; it is crucial their medical devices produce retinal images trusted for diagnosis. Currently, image quality parameters are measured to catch any devices failing standards on the production line, but the root cause and varying nature of several of these failures is not fully understood. Systems that involve scanning, such as Optos Laser Scanning Ophthalmoscopes, are subject to image quality issues in their manufacturing. Such issues arise through a combination tolerances of the electrical, optical and scanning components that make up the final product. In this project, the student will develop photonics/optics test rigs and algorithms for detecting artefacts in images and predicting root causes. It is not trivial to deliver an automated detection system that catches image artefacts. The designs of Optos' imaging systems change over time as a result of cost optimisations, parts obsolescence, new features, changes in supplied modules, and major new designs. By developing a test system that can adapt itself by learning from supplied examples, such as the algorithm proposed here, the improved algorithm-based quality assurance process can be run continuously and reduce the time required.