Wireless MRI techniques¹ could be useful in real-time image-guided surgery, by overcoming challenges of limited accessibility, electrical routing and cable interference. Previously, we demonstrated the concept of wirelessly powering implantable devices utilizing the B1 transmit field synchronously with imaging^2,3. Here, we develop system models for wireless powering using different MRI pulse sequences and compare wireless powering performance.Fig 1 shows the proposed concept and Fig 2 shows the developed system model. When the transmit pulse is active (modeled as an ideal switch in Fig 2), RF power is coupled into the harvester coil (modeled by a gain element) and rectified^2,3. A large output capacitor serves as an energy reservoir. The receive (or instrument activation switch) controls when power is supplied to the instrument or sensor being powered. A low-dropout output regulator, ensures that a constant ripple-free DC voltage is supplied. For the load, we consider a low-power Colpitts VCO in the medical radio frequency band of 403.55 MHz, driving a miniaturized antenna. The VCO was designed to operate at a bias point of 1.8 V and 2 mA (3.6 mW). providing an output power > -10 dBm to the antenna. Circuit level simulation models for each system component was developed using Agilent ADS 2014. Manufacturer-supplied models for the Schottky diodes and voltage regulators were used. Ideal switch models and transfer function-based gain elements were used to model the transmit amplification chain and the wireless coupling. The transfer function-based approximations can be justified by the relative homogeneity of the B1+ excitation field. To establish consistency of our model we performed measurements of wireless coupling of power (peak values) into a series-tuned 2 cm tuned rectangular harvester loop (on a 1.2 cm wooden dowel) when scanning using a fast SPGR sequence in a 3 T GE scanner. Peak RF energy coupled into the harvester coil was measured using a Rohde and Schwarz Spectrum Analyzer with a 40 dB attenuator in series (to protect the instrument from high instantaneous coupled power). System-level simulations using two candidate basic MRI pulse sequences were performed. Fast GRE sequences with extremely short TR can be implemented resulting in a greater frequency of transmit pulses. Fast spin echo uses several RF pulses of high flip angle (180) in a single TR, also resulting in a higher duty cycle of RF excitation pulses in a single TR. The time (in a single TR) and duty cycle for which the VCO-based RF transmitter could operate at different operating powers was simulated to analyze the performance capability of a data transponder powered simultaneously during imaging.

$$E=\int_{0}^{pulse length}C_{storage}V_{harvested}(t)\frac{ dV_{harvested}(t)}{dt} (1) $$

Fig 3 shows agreement between experimental and simulated peak coupled RF power into the harvester coil during imaging when acquiring axial fast SPGR images (Fig 4). Fig 4 shows some interference when the harvester coil is placed at the imaging slice. However, the artifact is significantly diminished in the slice just outside the harvester coil. The simulation results comparing different pulse sequences for wireless powering establishes that energy harvesting from the B1 field during MRI is a possibility. The sampling window during MRI signal acquisition is often less than 10% of TR and it is seen that the duty cycle is greater in most cases. This implies that imaging synchronous operation is realistic. Table I also gives some insight into choosing of scanning pulse sequences for wirelessly powering devices inside the scanner. fGRE pulse sequences using FA > 30° can potentially be operated throughout a T1-wighted scan. Hamming-windowed sinc pulses provide higher energy harvesting efficiency compared to rectangular pulses with same parameters. The relation of FA and duty cycle of the VCO is highly nonlinear as seen by difference in performance at 15° and 30° FAs. The required supply voltage could not be achieved by the rectifier circuit for the 15° rectangular pulse. This resulted in a 0% duty cycle. Use of a VCO as the driven load entails very specific bias points for consistent operation. The FSE pulse sequence is also a good candidate due to the repetitive 180° pulses. However, simulations require a large amount of memory and so were only performed between successive 180° pulses.

Experimental measurements of peak power coupled into the harvester coil agreed well with our predictions using the gain block. Since we relied on manufacturer supplied simulation models for most components, the simulation results are representative of the actual system. Insight gained into suitability of pulse sequences and scanning parameters for wireless powering is useful in pursing further system optimization and practical realization.