Estimation of Bladder Pressure from Pudendal Electrical Activity of Rabbit Based on the Self-Organized Recurrent Neural Network
Estimation of Bladder Pressure from Pudendal Electrical Activity of Rabbit Based on the Self-Organized Recurrent Neural Network
Meisam Baradaran,1Hamidreza Kobravi,2Ali Moghimi,3Saleh Lashkari,4,*
1. Department of Biomedical Engineering, Mashhad Branch, Islamic Azad University, Mashhad, Iran 2. Department of Biomedical Engineering, Mashhad Branch, Islamic Azad University, Mashhad, Iran 3. Rayan Center for Neuroscience & Behavior, Department of Biology, Faculty of Science, Ferdowsi University of Mashhad, Mashhad, Iran
Introduction: Hyper-reflexive and loss of voluntary control is a common result of neurological diseases such as spinal cord injury. Electrical stimulation is an alternative treatment for hyperreflexia. To inhibit the bladder, we need to detect nascent bladder contraction. Previous studies reported that electrical activity of the pudendal nerve is correlated with bladder pressure. The aim of this study is developing a neural network model for estimating bladder pressure during bladder contraction based on the electrical activity of the pudendal nerve.
Methods: Three models of neural networks were used to identify the relationship between bladder pressure and pudendal nerve activity: NARX, GRNN, and SRBFNN. Ten adult male rabbits were used in this study. Five experiments were performed to record bladder pressure at different conditions. Finally, to evaluate the performance of the ENG -pressure neural network model, the normalized root-mean-square (NRMS) index was calculated.
Results: NRMS for NARX, GRNN, and SRBFNN were obtained 12.16, 10.49, and 9.42, respectively. Results indicate that the SRBFNN network has less NRMS error than the other networks. Results shows proposed research is a promising approach to detect nascent bladder contraction.
Conclusion: Using recurrent neural network can characterize nascent bladder contraction