Channel modelling and localisation of capsule endoscope inside the small intestine

Capsule endoscopy is a relatively new gastrointestinal imaging procedure which has been used to inspect and diagnose possible abnormalities and diseases inside the human body. However one of the main drawbacks of this new technology is associating the exact location information of the capsule with the received video images to provide a clear idea of the location of any abnormalities. In this thesis we address different challenges associated with location estimation of a capsule endoscope. First we propose a radar system as a location estimation technique and a Received Signal Strength (RSS) method for capsule localisation. We also develop a new deterministic in-body path-loss model based on the different physical properties of tissues and with several reflections and absorptions by each tissue in the abdomen region. We show that a deterministic path-loss model which is only dependent on a theoretical analysis of radio propagation inside the abdomen area cannot fully ensure the accuracy of the model. A statistical in-body path-loss model is derived based on simulating the electromagnetic wave propagation inside the abdomen region of three distinct human phantom models. Based on the developed path-loss model the 2D location of the capsule is estimated using the trilateration method as well as the Nonlinear Least Squares (NLLS) algorithm. It is shown that the probability of achieving a location error of less than 15 mm is about 80% in a condition where the noise standard deviation is less than 8 dB. Moreover, using our developed statistical path-loss model and under three shadowing scenarios, we calculate the mathematical bound on the localisation precision. From the obtained results it can be concluded that a 2.4 GHz RSS-based localisation using a radar system is feasible and can reach centimetre-order precision.