Estimation of Inertial Measurement Unit Locations for Neuroimaging Applications
Date
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
Functional near infrared spectroscopy (fNIRS) is a portable neuroimaging modality that uses light to measure hemodynamic responses in the cortex of the brain. Knowing the exact locations of fNIRS optodes relative to each other is critical, because the distance between optodes directly affects the imaging depth. The status quo solution to optode localization is using third-party digitizers; however, they are expensive, time consuming, and defeat the portability of fNIRS. We propose embedding 9-axis inertial measurement units (IMUs) in optode hardware and developing an algorithm that can approximate the locations of IMUs relative to each other to quicken and cheapen this location estimation process. Said algorithm takes advantage of the head being an approximately rigid body and the difference in rotational accelerations experienced at different locations on a head to estimate IMU locations, growing more accurate as time passes. In preliminary recordings, the algorithm achieved an error of 5.94 ± 2.98 mm. While this error is inferior to a digitizer, the algorithm achieves this in 1/20th of the time and at 1/5th of the price of a digitizer. Since this solution is intended to be incorporated into optode hardware, it does not restrict the portable aspects of fNIRS. Further refinement to the algorithm via signal conditioning of accelerometer and magnetometer data as well as introducing extended Kalman filtering should significantly reduce error and is being actively investigated.