High-power switching and control system for concurrent TMS and MRI (under construction...)
Overview
We developed a switching and control system that allows the use of one coil for operation in both TMS and MRI modes. A switching box toggles between the two operation modes and the control system monitors the safety and transmits switching commands to the box itself.
Design
Two DCNEVT400 contactors (1.8kV, 400A) and two AEVT500 contactors (1.8kV, 3.3kA) are inside the switching box to switch in and out from MRI and TMS modes.
The Red Pitaya (FPGA, TEMLabs) and host computer coordinate operation mode transitions, MRI encoding sequences, and TMS pulse triggering.
The system reads an external trigger and starts the pre-programmed commands in the host computer. Through red pitaya, DAC, and control board, the command is executed as (1) encoding sequences that can be inputed into a gradient amplifier, (2) switching between different operation mode that instructs which pairs of contactors should be open/closed, and (3) TMS trigger that can be inputed into the TMS stimulator as the "trigger in" for firing TMS pulse.
Five fault conditions are actively monitored: (1) bad command, (2) TMS stuck open, (3) TMS stuck closed, (4) GPA stuck open, and (5) GPA stuck closed. Bad commands prevent unintended TMS pulses from being triggered when the TMS contactors are not fully closed or when a nonzero gradient amplifier current is detected during TMS mode. The TMS stuck open/closed and GPA stuck open/closed mechanisms continuously monitor the actual state of their respective contactors. If any fault condition is detected, the fault LED is activated, all contactors are immediately disengaged, and the system enters a safe shutdown state.
Circuit Design/Specifications
Click here to download switch box circuit design and related details: [[ https://rflab.martinos.org/images/9/95/Design.zip]]
Control
The Red Pitaya, the heart of the control section, has an AMD/Xilinx Zynq-7-series FPGA, which contains both programmable logic and a microprocessor running a Linux operating system from an SD card. The pulse timing sequence is represented by a CSV text file stored on the SD card, with a column for each control signal, and the rows representing the state of each signal at fixed 0.5 ms time steps. These control signals are: - Enable (digital) - TMS or GPA mode (digital) - Fire TMS pulse (digital) - GPA current setpoint (analog) Custom software running on the Red Pitaya’s microprocessor reads this pulse sequence file, and waits for an external trigger to run the sequence. When this external trigger arrives, the software uses a SPI protocol interface in the FPGA’s programmable logic to “re-play” the sequence of output signals on the DAC board. Click here for an example script that we used: [[ https://rflab.martinos.org/images/6/66/2_sequence_2sinthentms3_5vpk.txt ]]
Tests and Results
Lab bench tests were conducted to validate whether the system will safely switch between different TMS pulse strength and graident amplifier power level. For that, we tested the system with TMS strength of 10% and 20% combined with graident waveform current pk-pk of 6A, 12A, 18A, 30A.
Scanner tests were conducted to show if the system would induce local diffusion weighting without introducing any complications. We tested five scenarios inside a 3T scanner (SIEMENS MAGNETOM 3.0T): Scenario 1. Scanner baseline, Scenario 2. Encoding only, Scenario 3. TMS + Switching, Scenario 4. Encoding + Switching, and Scenario 5. Encoding + TMS + Switching. Imaging was captured using a spherical water phantom with the same spin echo EPI diffusion weighted sequence played out by 3T scanner (TE:50ms, TR:20000ms, Voxel Size: 2.0×2.0×6.0 mm³, 20 slices).
