Electronics‎ > ‎

Wearable accelerometer system

At the end of undergraduate studies in Poland, I was fortunate to intern at a physiological research laboratory, where I undertook a hands-on thesis project - a system for recording the triaxial accelerations of a wearer and displaying these in a GUI. Surprisingly, this mixture of electronics and software worked and was well-received. A brief description, along with the final document (English beginning page 4) and attachments are available here. Anyone interested has blanket permission to use or build upon them, with attribution to this page.

Summary

Parkinson's disease is a neurodegenerative condition responsible for major motor and cognitive dysfunction. Its major symptoms include a characteristic 3–7 Hz resting tremor, present particularly in the limbs. The magnitude of this tremor and the frequency of its occurrence can be used to assess the disease's progression and the effectiveness of current treatments. This thesis describes the design, construction, and programming of a compact, rechargeable acceleration recorder to support the ongoing research of the Mossakowski Medical Research Centre's Department of Applied Physiology (Polish Academy of Sciences).

The device consists of a microcontroller, accelerometer, SD card, indicator LED, rechargeable battery, and associated minor components. Attention was first devoted to the electrical design, wherein various components were supplied with appropriate, filtered voltages. A printed circuit board was designed, etched, and assembled. 3-D models of the device and casing were used to ensure that the device would be spatially effective, and the result is a tightly packed and durable arrangement. It is enclosed in a plastic casing with adjustable straps, which allow the device to be attached to a limb of interest and worn during everyday and diagnostic activities.

While not optimized for power consumption, the device can function throughout the day, and be chargedy using the accompanying charger and cable. The microcontroller was programmed to read the accelerometer's three-channel output via ADC and record it to the SD card often enough to capture not only the low frequencies of Parkinsonian tremor, but also higher physiological-tremor frequencies. Despite this, the card can store nearly three weeks' worth of continuous recordings. If the patient removes the card when plugging the device in, they could potentially carry it around for a month without needing to visit the researchers to download the data.

When the SD card is removed, the microcontroller enters a power-saving sleep mode. The
card's data can be transferred to a PC, where it can be read and processed by the accompanying MATLAB program. The program consists of a graphical user interface offering a wide variety of options for displaying time-acceleration and time-frequency data. It provides statistics on selected data and employs basic thresholds to detect and indicate the occurrence of higher-power spectral densities in selected frequency ranges. The latter can be used to detect the occurrence of Parkinsonian tremors. These could potentially be used to identify Parkinsonian tremors as well as other movements.
ċ
3D_layout_files.zip
(1462k)
Martin Berka,
May 15, 2017, 8:34 PM
ċ
EaglePCB.brd
(127k)
Martin Berka,
May 15, 2017, 8:32 PM
ċ
EagleSchematic.sch
(930k)
Martin Berka,
May 15, 2017, 8:32 PM
Ċ
Martin Berka,
May 15, 2017, 8:41 PM
ċ
Matlab_GUI_source_files.zip
(14k)
Martin Berka,
May 15, 2017, 8:36 PM
ċ
device_software_and_sample_data.zip
(3583k)
Martin Berka,
May 15, 2017, 8:36 PM