Performance Enhancement of an Achalasia Automatic Detection System Using Ensemble Empirical Mode Decomposition Denoising Method

Purpose: High-resolution manometry, a non-invasive tool for examination of esophagus, has been utilized in order for automatic detection of achalasia as one of the most recognized disorders of digestive system especially in the esophagus. Before designing any digestive system malfunction detection technique, it is necessary to pre-process the original recorded esophageal manometry signals in order to reject noise for further analysis since it is affected by a considerable amount of artifacts with different origins. Methods: Employing ensemble empirical mode decomposition for noise rejection, an artificial neural network is applied to detect achalasia automatically and classify the multi-channel esophageal manometry signal. Results: Results of this study show that the raw esophageal manometry signal requires a filtering step in order to provide accurate assessments. This approach is able to increase the accuracy of disorder diagnosis by more than 7% that is a quite significant amount. The accuracy of achalasia diagnosis in this study is 96.6%, which corresponds well in comparison with the other similar studies in the literature. Conclusion: It is highly recommended to attenuate the noise by a reliable technique such as the ensemble empirical mode decomposition method as a pre-processing step.