Despite advances in medical sciences, many patients cannot benefit from them due to the lack of resources, especially health care specialists in areas such as, in our case, vision diagnosis. The goal of this thesis is to develop an automated system to identify vision disorders, so that potential problems can be addressed as early as possible by having the system refer patients to a specialist without requiring extensive operator training or patient cooperation. This thesis explores the application of artificial neural networks and decision tree learning algorithms for diagnosing vision disorders by examining video images of patients' eyes. After employing a rigorous ten-fold testing methodology, the results indicate that the best system uses a decision tree approach and has an accuracy of 77% when evaluated against a specialist-recommended referral decision. Although these results do not outperform other reported research, the proposed approach has the advantage of requiring minimal cooperation to identify early signs of vision disorders.
展开▼