A Neural Based System For The Recognition of Gastrointestinal Diseases

Computer techniques can be applied to an incredibly wide range of problems in our life. Neural Network techniques give a hand in too many branches, and one of these branches is the diagnosis of diseases. In our research we have used a robust neural based technique to diagnose some of the Gastrointestinal diseases such as : Gastric Ulcer (G.U.), Duodenal Ulcer (D.U.), and Hiatus Hernia (H.H.). The patient is subjected to the measurement of electrogastric activity by placing electrodes on the outer skin of the abdomen which is a very comfortable technique, and a non invasive one as well. The system has passed several stages of tests and developments until a satisfactory performance has been achieved. The efficiency of the network has been enhanced using noise addition to escape the local minimum problem. A total of 93% of the diseases, reported by specialists in the field of gastrointestinal disease diagnoses, have been detected as definite with no false. Classification of diseases have been done such that high recognition rates are maintained while eliminating false detection between normal and abnormal cases.