Capsule Endoscope Localization Based On Embedded System

The birth of wireless capsule endoscopy brings the gospel for gastrointestinal patients, which canbe painless, non-invasive, and disposable to get clear pictures of the gastrointestinal tract. Thelocation information of the capsule endoscope can greatly help doctor’s clinical diagnosis, as wellas providing an important position and orientation parameters for the three-dimensionalreconstruction for target tissues. However, the existing technique can not accurately obtain thelocation and orientation of the wireless capsule endoscope.Here, we propose the static magnetic localization technique to track the position and orientationinformation of wireless capsule endoscope. The magnetic localization system collects magneticsignal from the permanent magnet inside the capsule through a three-axis magnetic sensor array,then the weak sensor signals are amplified and analyzed for further localization. In order toimprove the portability and flexibility of the system, we intend to design wearable magneticlocalization systems based on the embedded system.To determine the position and orientation of the magnet, an appropriate non-linear algorithm isdesigned based on the magnetic dipole model. It combines the PSO and the LM algorithm, inwhich the initial parameters are calculated by PSO and then LM algorithm calculates the finalparameters. So the algorithm can obtain good efficiency and accuracy.On the basis of the localization algorithm, we built a wearable magnetic localization modelsystem to overcome the shortcomings that patients can not move freely in the practical application.Sensor signals are captured by the sensor array which is suited to the human body. So the structureand location of the sensor distribution has a critical influence on accuracy and stability of the entiresystem. The simulation experiments have been implemented to optimize the structure of themagnetic sensors around the body.The accuracy of magnetic localization system is determined by the three aspects: theperformance of the magnetic sensors; the processing for the weak voltage signal; and the accuracyof localization algorithm. In addition, each magnetic sensor may have different parametersincluding sensitivity, position and direction. To improve the accuracy of the system, the iterativealgorithm has be applied to calibrate these parameters. Finally, we complete the magneticlocalization system by system calibration and the human-computer interaction software. Theexperiment results show that the system has the average position error0.9mm and the averageorientation error1.8degree.