Bayesian inference-based tracking for wireless capsule endoscopes

Wireless capsule endoscopy (WCE) has emerged as a convenient diagnostic method for human gastrointestinal (GI) diseases owing to its non-invasiveness and capability to explore the entire GI tract. It also has a large potential to play a therapeutic role owing to the rapid advances in micro-electromechanical systems (MEMS) technology. For accurate diagnosis and treatment of pathological conditions, a low-cost and accurate tracking system for WCE is highly required. Currently, the received signal strength (RSS)-based techniques are widely used for WCE localization because of its advantages in terms of non-specificity and low-cost implementation. However, these RSS-based techniques are quite susceptible to RSS measurement noise with random characteristics. We develop the Bayesian graphical model (BGM) for the RSS-based tracking system and then use Gibbs sampling to stochastically infer the location of the capsule endoscope. Through the results of the simulation experiment, we demonstrate the validity of the proposed methodology for WCE-tracking system.