Magnetic bandages for 3D limb scanning
Bandages fitted with magnetic sensors that capture the 3D shape of feet and residual limbs for orthotists and prosthetists.
- 1Seven bandages wrapped around the limb, three on the foot (one around the heel) and four up to the knee, each carrying about a hundred MMC5603 magnetic sensors.
- 2Control electronics on each bandage, next to the velcro fastening: an nRF52832 handles BLE and collects the measurements over I²C from ATtiny microcontrollers that drive the MMC5603 sensors.
- 3Foldable field source with six coils, on a tripod along the leg. Its nRF52840 master switches the coils on one after the other, plus an idle phase to subtract the Earth's field, and drives the bandages over BLE.
- 4Desktop application, connected to the source over USB: pairing, acquisition start and GPU-accelerated multilateration.
- 5Sparse point cloud: dense where the bandages are, empty in between.
- 6Dense point cloud: an AI model upsamples the cloud and fills in the missing areas.
- 7STL mesh, ready for the orthotist or prosthetist to design the device.
- Category
- Medtech & research
- Organization
- FHNW
- Period
- 2023 – 2026
- Technologies
- nRF52840, nRF52832, ATtiny, MMC5603, Zephyr, BLE, USB, GPU, C++, Altium
- Website
- bellwaldtec.ch ↗
Context
Taking the shape of a foot or a residual limb usually means a plaster cast and keeping the patient still, which is difficult with children or spastic patients. Here, the practitioner wraps seven bandages fitted with magnetic sensors around the limb and places a foldable field source on a tripod next to it. The source switches its coils on one after the other while the sensors measure the field, and the whole acquisition takes 2 seconds. A multilateration computation turns the measurements into a sparse point cloud, an AI model fills in the gaps between the bandages, and the dense cloud becomes an STL model used to design the orthosis or prosthesis. The practitioner's hands do not disturb the measurement, so they can hold and correct the limb during the scan. The technology was developed at the FHNW and is being brought to market by the spin-off BellwaldTEC.
My role
Led the firmware architecture and implementation for real-time data acquisition: an nRF52840 master on the field source, connected to the PC over USB, and an nRF52832 on each of the seven bandages, reading ATtiny sensor controllers over I²C, low-power design, pairing, deterministic sequencing of the coil phases and BLE synchronisation of the sensing devices. Developed the USB link to the PC and the GPU implementation of the multilateration, which brings the computation time down drastically, and the C++ test software used as a development tool to debug, test and develop the system. Designed the flexible and rigid PCBs and maintained the technical documentation.