Previous definitions of “wellness,” limited to a brisk walk and eating a few more vegetables, are in many ways like the distant past. Shiny watches and stylish rings now measure how we eat, sleep and breathe, using a combination of motion sensors and microprocessors to analyze bytes and bits.
Even with the current variety of smart jewelry, clothing, and temporary tattoos that seem both complex and manageable, scientists at MIT’s Laboratory for Computing and Artificial Intelligence (CSAIL) and the Center for Artificial Intelligence (CPAI) of Massachusetts General Hospital (MGH) wanted to do things a little more personal. They created a toolkit for designing health and motion detection devices using what’s known as “electrical impedance tomography (EIT),” a fancy word for an imaging technique that measures and visualizes the internal conductivity of a person. (EIT is typically used for things like observing lung function or detecting cancer.)
Using the ‘EIT kit’, the team designed a range of devices supporting different sensing applications, including a personal muscle monitor for physical rehabilitation, a portable hand gesture recognition device and a “Bracelet” capable of detecting distracted driving.
While EIT detection typically requires expensive hardware setups and complicated image reconstruction algorithms, the use of printed electronics and open source EIT image libraries has made it an attractive, inexpensive, and portable option. But designing EIT articles remains difficult and usually requires a good fusion of design knowledge, adequate device-human contact, and optimization.
The EIT-kit 3D editor places the user in the driver’s seat for full design direction. Once the sensing electrodes (which measure human activity) are placed on the device in the editor, they can be exported to a 3D printer. The item can then be assembled, placed on the target measurement area, and connected to the EIT kit sensing motherboard. Finally, an on-board microcontroller library automates electrical impedance measurement and allows users to view measured data visually, even on a mobile phone.
EIT-Kit is a toolkit for designing wearable devices that use electrical conductivity to detect movement and monitor health.
Existing devices can also only detect movement, limiting users to understanding only how they change position over time – but the EIT kit can detect actual muscle activity. A device prototyped by the team, which looks like two single bands, detected muscle strain and tension in the thigh to monitor muscle recovery after injury, and can even be used to prevent re-injury. The muscle monitor here used two arrays of electrodes to create a 3D image of the thigh, as well as augmented reality to visualize muscle activity in real time. In this case, the simple detection of movement would be useless, since a person performing a rehabilitation exercise correctly requires using the correct muscle. In addition, the researchers detected biological tissue via an EIT device that analyzed the tenderness of raw meat.
“The EIT-kit project fits with my long-term vision of creating personal health sensing devices with rapid function prototyping techniques and new sensing technologies,” said Junyi Zhu, PhD student at MIT CSAIL, senior author of a new article on the EIT kit. “During our study alongside healthcare professionals, we found that EIT detection is highly dependent on the patient and location of detection, due to measurement setups, signal calibration, placement of electrodes and other bioelectrical factors. These challenges can be solved with customizable hardware and automation algorithms. Beyond EIT, other health sensing technologies face similar complexities and personalized needs. ”
The team is currently working with MGH to develop an EIT kit to create remote rehabilitation devices to monitor different parts of a patient’s body during healing. Since all devices in the EIT kit are mobile and personalized to suit the patient’s body shape and injury, they can be easily used at home to give doctors a more holistic picture of the healing process.
Zhu wrote the article alongside Jackson Snowden, an MIT alumnus, alumni Joshua Verdejo ’21, MNG ’21 and Emily Chen ’21, CSAIL doctoral student Paul Zhang, CPAI co-director at MGH and the Harvard Medical School instructor Hamid Ghaednia, head of spine surgery at MGH and Harvard Medical School associate professor Joseph H. Schwab, and MIT associate professor Stefanie Mueller.
This material is based on work supported by the National Science Foundation. The project was carried out in collaboration with Schwab and Ghaednia. It will be showcased at the 2021 ACM Software and User Interface Technologies (UIST) Symposium next month.