Analysis of EGG Signals for Digestive System Disorders Using Neural Networks

Abstract

A novel approach is proposed to analyze the digestive system disorders using the developed Electrogastrogram [EGG] setup. Current days the role of Endoscopy is more for the investigation of digestive system disorders, which is very tedious, expensive and invasive. EGG is a non-invasive, cheap and painless study. EGG is considered as a preliminary investigation before the Endoscopy procedure. Based on the finding from the EGG, if it is uncomplicated Gastric diseases, Benign (non cancer) we can start the treatment solely on EGG report alone. Only in case of suspected Malignancy (cancer) and complicated benign disorders we can go for Endoscopy procedure. The biosignal are recorded from the patients and normal individual cutaneously from the stomach includes the various stages such as amplification, filtering, pulse width modulation and demodulation. The EGG data are acquired using DSO or Data scope software and the same is analyzed using Matlab’s neural network toolbox. The Adaptive Resonance Theory (ART1) and Learning vector Quantization (LVQ) networks are used as a Neural Network classifiers to classify the digestive system disorders. The proposed method is used to investigate the digestive system disorders such us ulcer, dyspepsia, etc.The investigation is carried out with 180 human being includes inpatient, outpatient of gastroenterology department and normal individual and normal individual. As a result of this investigation the EGG values for ulcer patient is between 4–4.5 cpm (cycle per minute), dyspepsia patient is between 1–2.5 cpm and for normal individual is between 3–3.5 cpm.The proposed new method results in greater accuracy comparable with other conventional methods to provide fair amount of accuracy.