Knowledge acquisition in expert system assisted diagnosis a machine learning approach

Two-dimensional gel electrophoresis is a biochemical technique for protein separation, producing complex images containing thousands of spots. Their expert system assisted interpretation would be of great use in a clinical environment. Nevertheless, there are currently no experts capable of analysing these images. A machine learning system is presented that allows the expert system to acquire new knowledge and to formulate rules to diagnose new diseases from such a picture. A heuristic clustering method is used to group the images into separate classes and to determine the typical spots for each disease.