This is a collection of work done to create a multi-use MRI compatible wireless sensor system using off the shelf components. This example shows how to simultaneously collect information from two different sensors located inside of an MRI bore.
- Raspberry Pi 4
- Adafruit ESP32 Feather board (x2)
- ESP32 U.FL module
- u.fl PCB antenna
- Sparkfun ADXL313 Accelerometer breakout board
- Adafruit BNO08x IMU breakout board
- Nickel-free Lithium-ion battery
- hookup wire
- solder
- copper tape/metal box
The Raspberry Pi acts as a center to collect and save data from all of the different wireless sensor nodes. Each of the feathers can be connected to a number of different sensors, limited by whatever protocol the sensor needs in order to work, whether that be GPIO pins with SPI, or intersections with I2C, or clock speed depending on your desired sampling rate.
There's a few modules that are needed on the Pi to function. Below are instructions on how to install the modules, and the code for the NodeRED flow is in the repository.
To act as an independent wireless hub for the sensors, the Raspberry Pi can act as a standalone Wireless Access Point (WAP). This means it acts like a Wi-Fi router, letting multiple devices connect to its network and share information with each other. There are several ways to do this, I have used RaspAP to get started. I would highly recommend to change the default password for the RaspAP module.
To organize the communication of information between the Raspberry Pi and the sensors, the Raspberry Pi ran as an MQTT broker using Mosquitto. There are several tutorials on how to set this up already, I have used the one from randomnerdtutorials. I used the default settings.
The last thing needed is a way to interface with the devices in order to see what is connected and to save information. Again, several ways to do this, but I ended up using NodeRED to take care of a lot of the heavy lifting for me. It also allow for custom scripting with JavaScript when the provided nodes do not suit your needs. You need to also install the dashboard community module (I used v 2.20.0) to use the flow I have provided here to be an interface to collecting data from the sensors.
A nickel free Li-ion battery (from Powerstream) is used to power an Adafruit ESP32 Feather which is connected to an Adafrut BNO08x breakout board over I2C. Four wires are needed between the Feather and BNO08x (VDD, GND, SCL, SDA). I modified the default Adafruit ESP32 Feather to have an u.FL version of the ESP32 module. This is to allow for use of an external u.FL Wi-Fi antenna, to allow the rest of the sensor node be encapsulated by a metal box to prevent MRI interference.
All of the hardware except the antenna can be encapsulatd within a copper box to prevent interference. Note that depending on the components you choose to use there is a chance they will heat up in such an enclosed space. Do safety temperature tests with each change you make.
I would recommend using the PlatformIO (I use it as an extension of Visual Studio Code) to manage dependencies automatically. The main.cpp file should be able to be used as a base to be able to compiile and upload using the Arduino IDE. The following libraries are needed, and defined in the platform.ini file.
- knolleary/PubSubClient@^2.8
- bblanchon/ArduinoJson@^6.17.3
- adafruit/Adafruit BNO08x @ ^1.2.1
The PubSub client is used to talk to the Raspberry Pi's MQTT broker. The Arduino JSON library is used to tidily pack up the sensor information before transferring the information over MQTT. The Adafruit library is used to talk with the BNO08x IMU, accessing the different sort of information that the IMU provides, such as linear acceleration, raw acceleration, or orientation.
The Visual Studio Code version I used is 1.63.1. The PlatformIO version I used is Core 5.2.3 and Home 3.4.0.
-
Connect sensor to the microcontroller.
-
Upload proper firmware to the controller
- If running a new sensor with custom code, I would run a simple test sketch from a tutorial to ensure that microcontroller<->sensor communication is happening properly before adding the overhead of the wireless. Serial monitor is your friend.
-
Plug in Raspberry Pi
-
Wait a few minutes
-
Connect to the raspi-webgui Wireless network. The computer will yell at you that there is no internet, but will connect to the network.
-
Open a browser (eg, Firefox) and go to 10.10.10.10:1880/ui, a UI similar to this should pop up
-
Before powering on the sensors, the Raspberry Pi time needs to be adjusted!
- Open a ssh client of your choice. I like MobaXTerm because it also allows for easy file transfer
- SSH into 10.10.10.10 and log in with Pi's username/pw
- Run the following command (adjust the information as necessary)
- sudo date -s "Wed Nov 17 20:57:00 EDT 2021"
- I usually open up the computer clock to get the seconds as close as possible, having the command be for the next minute and hitting enter in the ssh terminal right when the laptop clock hits the correct time. It does not need to be exact, but it helps when plotting the data to see if the sensor is connected.
- Open a ssh client of your choice. I like MobaXTerm because it also allows for easy file transfer
-
Now that the time is correct, the graphs should play nicely!
-
Now you can enable the 'show sensor' part in the browser then power up one of the microcontroller boards. You should now see a graph that shows the x-axis of the attached accelerometer.
-
In order to save information, you need to fill out the file name information in the form in the user interface and hit submit. The folder/filename should appear above the save icon. When you are ready to save information from the sensors, you can hit the record toggle and information will start saving to the file. There is not a check to see if the file already exists, but will simply keep appending information to the filename.
Have a magnet on hand to test the magnetic properties of components before bringing close to the MRI machine. If using a small enough magnet, you can isolate which component needs to get removed/replaced and find an appropriate replacement, such as the USB connector or battery. Not all parts will be able to be replaced. The Adafruit ATWINC1500 board could not be used in a 3T magnet due to the magnetic properties of the shield and components in the wireless module. Capacitors and USB connectors are commonly magnetic, but usually have non-magnetic components also available to order, but not as common.