Improved Water Strider Optimization with Deep Learning based Image Classification for Wireless Capsule Endoscopy

Abstract

Wireless capsule endoscopy (WCE) allows physicians to observe the digestive tract without doing surgery, at the cost of a huge volume of images should be analysed. The analysis and interpretation of WCE images will be a complicated task which needs computer aided decision (CAD) mechanism for assisting medical practitioner with the video screening and, lastly, with the diagnosis. Manual examination of WCE is a time taking process and can be benefitted from automatic detection by utilizing artificial intelligence (AI). Deep learning was a new method related to neural network. WCE was the criterion standard to identify small-bowel diseases. In this context, this study formulates an Improved Water Strider Optimization with Deep Learning Based Image Classification (IWSO-DLIC) for WCE. The presented IWSO-DLIC technique examines the WCE images for the identification of diseases. For image pre-processing, the IWSO-DLIC technique uses Wiener filtering (WF) approach. In addition, the IWSO-DLIC technique employs MobileNet feature extractor, and the hyperparameter tuning process takes place via the IWSO algorithm. Moreover, the IWSO algorithm is designed by the combination of oppositional based learning (OBL) concept with standard WSO algorithm. Finally, to classify WCE images, long short-term memory (LSTM) model is employed in this study. To demonstrate the enhanced performance of the IWSO-DLIC model, a series of simulations were performed. The simulation values stated the enhanced performance of the IWSO-DLIC technique over other recent models.