Diagnosis of medical diseases is facing a challenge of knowledge discovery from the growing volume of data. Patient data sheet may be text (high level), images context (high level), and image content (low level) data. Data mining consists of extracting interesting knowledge from text data. Content and context of an image can mutually help infer and reinforce one another. By hybridizing data mining with image classification using content and context information, a new decision making method is obtained to solve this diagnosis problem. Content based image retrieval is used to support the decision possibility.
This paper presents an expert system for mitral valve diseases. This system provides the proper timing and indications of surgery in mitral valve diseases. The acquired knowledge is obtained from a hybrid combination of domain expert, medical data mining, context, and content based image retrieval. The system has updating and explanation units to update medical knowledge and answer the causes of diagnosis.
|
