Ultra-low-power Physical Activity Classifier for Wearables: From Generic MCUs to ASICs.

In the era of Internet of Things (IoT), an increasing amount of sensors is being integrated into intelligent wearable devices. These sensors have the potential to produce a large quantity of physiological data streams to be analyzed in order to produce meaningful and actionable information. An important part of this processing is usually located in the device itself and takes the form of embedded algorithms which are executed into the onboard microcontroller (MCU). As data processing algorithms have become more complex due to, in part, the disruption of machine learning, they are taking an increasing part of MCU time becoming one of the main driving factors in the energy budget of the overall embedded system. We propose to integrate such algorithms into dedicated low-power circuits making the power consumption of the processing part negligible to the overall system. We provide the results of several implementations of a pre-trained physical activity classifier used in smartwatches and wristbands. The algorithm combines signal processing for feature extraction and machine learning in the form of decision trees for physical activity classification. We show how an in-silicon implementation decreases up to 0.1 µW the power consumption compared to 73 µW on a general-purpose ARM's Cortex-M0 MCU.

Electrical Impedance to Assess Facial Nerve Proximity During Robotic Cochlear Implantation.

Reported studies pertaining to needle guidance suggest that tissue impedance available from neuromonitoring systems can be used to discriminate nerve tissue proximity. In this pilot study, the existence of a relationship between intraoperative electrical impedance and tissue density, estimated from computer tomography (CT) images, is evaluated in the mastoid bone of in vivo sheep. In five subjects, nine trajectories were drilled using an image-guided surgical robot. Per trajectory, five measurement points near the facial nerve were accessed and electrical impedance was measured (≤1 KHz) using a multipolar electrode probe. Micro-CT was used postoperatively to measure the distances from the drilled trajectories to the facial nerve. Tissue density was determined from coregistered preoperative CT images and, following sensitivity field modeling of the measuring tip, tissue resistivity was calculated. The relationship between impedance and density was determined for 29 trajectories passing or intersecting the facial nerve. A monotonic decrease in impedance magnitude was observed in all trajectories with a drill axis intersecting the facial nerve. Mean tissue densities intersecting with the facial nerve (971-1161 HU) were different (p <0.01) from those along safe trajectories passing the nerve (1194-1449 HU). However, mean resistivity values of trajectories intersecting the facial nerve (14-24 Ωm) were similar to those of safe passing trajectories (17-23 Ωm). The determined relationship between tissue density and electrical impedance during neuromonitoring of the facial nerve suggests that impedance spectroscopy may be used to increase the accuracy of tissue discrimination, and ultimately improve nerve safety distance assessment in the future.

A Neuromonitoring Approach to Facial Nerve Preservation During Image-guided Robotic Cochlear Implantation.

HYPOTHESIS: A multielectrode probe in combination with an optimized stimulation protocol could provide sufficient sensitivity and specificity to act as an effective safety mechanism for preservation of the facial nerve in case of an unsafe drill distance during image-guided cochlear implantation. BACKGROUND: A minimally invasive cochlear implantation is enabled by image-guided and robotic-assisted drilling of an access tunnel to the middle ear cavity. The approach requires the drill to pass at distances below 1  mm from the facial nerve and thus safety mechanisms for protecting this critical structure are required. Neuromonitoring is currently used to determine facial nerve proximity in mastoidectomy but lacks sensitivity and specificity necessaries to effectively distinguish the close distance ranges experienced in the minimally invasive approach, possibly because of current shunting of uninsulated stimulating drilling tools in the drill tunnel and because of nonoptimized stimulation parameters. To this end, we propose an advanced neuromonitoring approach using varying levels of stimulation parameters together with an integrated bipolar and monopolar stimulating probe. MATERIALS AND METHODS: An in vivo study (sheep model) was conducted in which measurements at specifically planned and navigated lateral distances from the facial nerve were performed to determine if specific sets of stimulation parameters in combination with the proposed neuromonitoring system could reliably detect an imminent collision with the facial nerve. For the accurate positioning of the neuromonitoring probe, a dedicated robotic system for image-guided cochlear implantation was used and drilling accuracy was corrected on postoperative microcomputed tomographic images. RESULTS: From 29 trajectories analyzed in five different subjects, a correlation between stimulus threshold and drill-to-facial nerve distance was found in trajectories colliding with the facial nerve (distance <0.1  mm). The shortest pulse duration that provided the highest linear correlation between stimulation intensity and drill-to-facial nerve distance was 250  µs. Only at low stimulus intensity values (≤0.3  mA) and with the bipolar configurations of the probe did the neuromonitoring system enable sufficient lateral specificity (>95%) at distances to the facial nerve below 0.5  mm. However, reduction in stimulus threshold to 0.3  mA or lower resulted in a decrease of facial nerve distance detection range below 0.1  mm (>95% sensitivity). Subsequent histopathology follow-up of three representative cases where the neuromonitoring system could reliably detect a collision with the facial nerve (distance <0.1  mm) revealed either mild or inexistent damage to the nerve fascicles. CONCLUSION: Our findings suggest that although no general correlation between facial nerve distance and stimulation threshold existed, possibly because of variances in patient-specific anatomy, correlations at very close distances to the facial nerve and high levels of specificity would enable a binary response warning system to be developed using the proposed probe at low stimulation currents.