The standard way to read muscle intent for a prosthetic limb is electromyography (EMG): skin electrodes picking up the electrical signal a muscle fires when it activates. It’s noninvasive and well-established, but it’s a proxy — it measures electrical activation, not what the muscle is actually doing mechanically, and that gap shows up as real problems: the signal fades when a muscle holds a fixed length, nearby muscles bleed into the reading, and sweat disrupts contact.
A new study from Hugh Herr’s Biomechatronics group at the MIT Media Lab, working with the K. Lisa Yang Center for Bionics, tests a different approach called magnetomicrometry directly in humans for the first time. Lead authors Christopher Shallal (MIT graduate student) and Cameron Taylor (PhD ‘20, now at UNC Chapel Hill), with plastic surgeon Matthew Carty of Brigham and Women’s Hospital, implanted pairs of small (roughly 3mm) biocompatible magnetic beads into the ankle muscles of three people with below-knee amputations — up to four bead pairs per participant. External sensors, worn over the skin, track the distance between each bead pair wirelessly from up to 6cm deep, which yields muscle length and velocity directly, rather than inferring activation electrically.
The comparison the study ran is the part that matters for anyone building prosthetic control interfaces: magnetic tracking against both implanted-electrode EMG and surface EMG, in a virtual cursor-control task. The magnetic approach produced more accurate and stable control than either EMG method. That’s a direct, mechanical measurement winning out over an electrical proxy for the thing you actually care about — what the muscle is doing, not just whether it fired.
The underlying magnet-pair technique was first demonstrated in animal studies in 2021; this is the extension into long-term human implants, published in Nature Sensors as “A wireless magnetic implant system for continuous neuromuscular sensing.” The authors frame it as less invasive than implanted electrodes while still delivering a direct mechanical readout — a middle point between surface EMG’s convenience and implanted EMG’s precision, with a different failure profile than either.
For anyone tracking input modalities beyond conventional EMG and hand-tracking (this site has covered both at length), magnetomicrometry is worth watching specifically because it measures something EMG structurally cannot: actual muscle mechanics, continuously, regardless of whether the muscle is actively changing length or holding still.