Using heterogeneous sensory measurements in a compliant magnetic localization system for medical intervention

In many medical intervention procedures, passive magnetic tracking technology has found favor in continuous localization of medical instruments and tools inside the human body. By utilizing a small permanent magnet as a passive source, it requires no dedicated power supply or wire connection into the body. Past researches usually adopt rigid structures to restrict the movement of sensors, as the precise positional information of the homogeneous magnetic sensors play an important role in the accuracy of traditional inverse optimization algorithms. In this paper, we investigate methods to enable the sensing system to be used for the nasogastric (NG) tube localization in a compliant setting, such that the device can conform around the patient for improved ergonomics and comfort. Such a system, which now contains additional sensors required to sense the active compliance, will contain a non-homogeneous sensor assembly producing heterogeneous sensory information. Two methods are proposed and evaluated: one is a modified inverse optimization method using a deformation model in series with the magnetic field model; the other is a direct forward Artificial Neural Network (ANN) method. The efficacy of both methods were evaluated and compared by numerical simulation and experiments. Advantages and disadvantages of both methods were discussed at the end.