Optimal Spatial Sensor Design for Magnetic Tracking in a Myokinetic Control Interface. (November 2021)

• Record Type: Journal Article
• Authors: Gherardini, Marta; Mannini, Andrea; Cipriani, Christian
• Is Part Of: Computer methods and programs in biomedicine. Volume 211(2021)
• Publication Date: 2021-11
• Subjects: Magnetic sensors -- Magnetic tracking -- Myokinetic control interface -- Sensor optimization -- Prosthetic hand; Medicine -- Computer programs -- Periodicals; Biology -- Computer programs -- Periodicals; Computers -- Periodicals; Medicine -- Periodicals; Médecine -- Logiciels -- Périodiques; Biologie -- Logiciels -- Périodiques; Biology -- Computer programs; Medicine -- Computer programs; Periodicals; Electronic journals; 610.28
• DOI: 10.1016/j.cmpb.2021.106407
• Languages: English
• ISSNs: 0169-2607

Abstract

Highlights: A strategy for optimizing the sensing system design for tracking multiple magnets in a 3D anatomical workspace. Results hold the potential of improving the efficiency of real-time control of upper-limb prostheses. The sensor selection strategy is intuitive and fast, and easily extendable to different settings and applications involving magnetic tracking. Abstract: Background and Objectives: Magnetic tracking involves the use of magnetic sensors to localize one or more magnetic objectives, in those applications in which a free line-of-sight between them and the operator is hampered. We applied this concept to prosthetic hands, which could be controlled by tracking permanent magnets implanted in the forearm muscles of amputees (the myokinetic control interface). Concerning the system design, the definition of a sensor distribution which maximizes the information, while minimizing the computational cost of localization, is still an open problem. We present a simple yet effective strategy to define an optimal sensor set for tracking multiple magnets, which we called the Peaks method. Methods: We simulated a proximal amputation using a 3D CAD model of a human forearm, and the implantation of 11 magnets in the residual muscles. The Peaks method was applied to select a subset of sensors from an initial grid of 480 elements. The approach involves setting an appropriate threshold to select those sensors associated with the peaks in the magnetic flux density and its gradient distributions. Selected sensors were used to track the magnets during muscle contraction. For validating our strategy, an alternative method based on state-of-the-art solutions was implemented. We finally proposed a calibration phase to customize the sensor distribution on the specific patient's anatomy. Results: 80 sensors were selected with the Peaks method, and 101 with the alternative one. A localization accuracy below 0.22 mm and 1.86° for position and orientation, respectively, was always achieved. Unlike alternative methods from the literature, neither iterative or analytical solution, nor a-priori knowledge on the magnet positions or trajectories were required, and yet the outcomes achieved with the two strategies proved statistically comparable. The calibration phase proved useful to adapt the sensors to the patient's stump and to increase the signal-to-noise ratio against intrinsic noise. Conclusions: We demonstrated an efficient and general solution for solving the design optimization problem (i.e. identifying an optimal sensor set) and reducing the computational cost of localization. The optimal sensor distribution mirrors the field shape traced by the magnets on the sensing surface, being an intuitive and fast way of achieving the same results of more complex and application-specific methods. Several applications in the (bio)medical field involving magnetic tracking will benefit from the outcomes of this work.