Localization and Tracking of Intestinal Paths for Wireless Capsule Endoscopy
MetadataShow full item record
Wireless capsule endoscopy (WCE) is a non-invasive technology used for visual inspection of the human gastrointestinal (GI) tract. Localization of the capsule is a vital component of the system, as this enables physicians to identify the position of abnormalities. Several approaches exist that use the received signal strength (RSS) of the radio frequency (RF) signals for localization. However, few of these utilize the sparseness of the signals. Due to intestinal motility, the capsule positions will change with time. The distance travelled by the capsule in the intestine, however, remains more or less constant with time. In this thesis, a compressive sensing (CS) based localization algorithm is presented, that utilize signal sparsity in the RSS measurements. Different L1-minimization algorithms are used to find the sparse location vector. The performance is evaluated by electromagnetic (EM) simulations performed on a human voxel model, using narrow-band (NB) and ultra wide-band (UWB) signals. From intestinal positions, the distance the capsule has travelled is estimated by use of Kalman- and particle filters. It was found that localization accuracy of a few millimeters is possible under ideal conditions, when the RSS measurements are generated from a path loss model. When using path loss data from the EM simulations, localization accuracy on the order of 20-30 mm was achievable for NB signals. Use of UWB signals resulted in localization errors between 35-60 mm, depending on frequency range and bandwidth. From generated intestinal positions, the travelled distance was estimated with a minimum accuracy of a few millimeters, when using a VNL Kalman filter and moderate amounts of observation noise. The results are found from a limited amount of data. In order to increase the confidence in the presented results, the performance of the localization algorithm and the filters should be evaluated with a larger number of datasets.