Cost-Efficient Real-Time Adaptive Location Tracking With Interacting Multiple Transition Model for Implantable Medical Device

Abstract

One of the main tasks connected with implantable medical devices, such as wireless capsule endoscope (WCE) is location tracking of a capsule moving inside a human body. To improve the localization accuracy in a real situation, a transition model to describe the capsule movement inside the body should be appropriately considered, where the trajectory segments are composed of different movement models. In this paper, in addition to the conventional random way point model, we constructed an interacting multiple model (IMM), which is a combination of two different movements to represent suitable WCE movement. Then, we proposed a combined extended Kalman and particle filter method with respect to the balance between the accuracy and computational cost for realizing precise real-time WCE location tracking. Compared with conventional tracking methods, the proposed tracking location method with the IMM transition model achieved 30% better estimation accuracy and 98% lower computational cost.