ISO/TS 10974¹ proposed a four-tier approach² which addresses the MRI-related risk from RF-induced heating of active implants. The effort required to demonstrate safety increases with the tiers, and the higher tiers (tiers 3&4) are known to be necessary for demonstrating safety of elongated implants, such as cardiac pacemakers and neurostimulators.

The tier 3 method of ISO/TS 19074 is designed to eliminate the computational requirements of multi-scale full-wave modeling required in Tier 4 (Fig 1), such that the RF-induced heating of leaded-implants may be assessed in practice while still maintaining a progressive margin to ensure patient safety. It requires a separate evaluation of the a) implant’s ability to pick up RF energy along its length and b) EM exposure of patients undergoing MRI examinations. The output of a) is a mathematical model that relates local absorption in the vicinity of the implant to the incident electric field along its length and the output of b) is the tangential component of the complex electric field distribution along clinical routings of the implant in patients.

The derivation of a) for elongated conductive structures (e.g., cardiac pacemakers or deep-brain stimulators) via numerical methods can be challenging or even unfeasible. The computational burden is dependent on the complexity of the implant structure (e.g. tightly-wound helical conductors). We present the development and validation of an experimental system which enables the RF characterization of leaded-implant regardless of its complexity. The predictive models of several generic implants are generated with the experimental system, and are validated against their numerically-derived counterparts.

Piece-wise excitation³ ($$$\pi$$$X) is a method used to characterize the implant's ability to pick up RF energy along its length. It is in compliance with the revised Tier 3 of ISO/TS 19074. Several studies^3-5 have shown the potential applicability of the method for predicting the local RF-heating of implants against experimental tests.

The $$$\pi$$$X method is briefly described here. A predictive model of an implant ($$$\pi$$$X model), $$$h(l)$$$, is defined as the relationship between the local induced electric field around a tip or an electrode pole of an implant and an excitation along length $$$l$$$ of the implant. The tangential component of the local incident electric field, $$$E_\textrm{tan}$$$, is coupled with the implant at length $$$l$$$ and the induced electric field around an electrode (or radiated components), at point $$${\bf r}$$$ is evaluated. The total induced electric field at point $$${\bf r}$$$ attributed to $$$E_\textrm{tan}$$$ coupling along the entire implant of length $$$L$$$ can be calculated from the relation:

$$E_{\textrm{induced},{\bf r}} = \int_0^L h(l)E_\textrm{tan}(l)dl$$

The $$$\pi$$$X method is realized experimentally and Fig 2 illustrates the $$$\pi$$$X system, developed by ZMT Zurich MedTech AG. Generic implants of length 400, 600, and 800-mm are characterized with the $$$\pi$$$X system (Fig 3). The implants are insulated 1.5 mm-diameter stainless steel wires, with insulation thickness of 0.5 mm. At one end of the wire, the insulation is removed over the length of 10 mm from the end of wire. The other end of the wire is capped with insulation and no metallic parts are exposed. All implants were characterized at 64 MHz and in homogeneous media simulating high-water content tissue ($$$\epsilon_r$$$ = 78, $$$\sigma$$$ = 0.47 S/m). The combined uncertainty budget of the $$$\pi$$$X system is assessed to be 1.06 dB and 9.6 degrees, in the measured magnitude and phase, respectively (Fig 2, right).

The experimentally-obtained models are compared against their numerical counterparts. The numerical models were derived from full-wave computational electromagnetics simulations using Sim4Life (ZMT Zurich MedTech AG) for the same homogeneous media and also at 64 MHz. The associated numerical uncertainty was estimated to be 0.2 dB.

Fig 4 compares the models obtained with the $$$\pi$$$X system to those from full-wave simulations. The results indicate maximum deviations from the numerical targets of up to 0.5 dB in magnitude and 8 degrees in phase; both are below the combined uncertainty of experimental and numerical evaluations. Therefore, the $$$\pi$$$X models are accurate within the established confidence interval.

The $$$\pi$$$X system operates within the assessed uncertainty budget, and the $$$\pi$$$X models derived with the system are accurate within the established confidence intervals. The $$$\pi$$$X model is derived under simplified in vitro conditions, and its applicability to in vivo context of use must be determined before the model can be used to obtain meaningful estimates of in vivo RF-heating in patients. The model validation method for in vivo applications is a topic for future investigation.