Abstract
The objective of this paper is to investigate the effect that lateral spatial resolution of ultrasound has on finger flexion classification. The objective's purpose leads toward the development of a wearable human machine interface (HMI) using single-element ultrasonic sensors with non-focused ultrasound. Ultrasound radiofrequency (RF) signals were acquired using a linear array ultrasound probe in B-mode while performing the individual finger flexions from five healthy volunteers. Each B-mode frame is composed of 127 parallel ultrasound RF signals along the lateral direction within a 40-mm width. To reduce the lateral resolution of ultrasound data artificially, the RF signals were averaged into a reduced number of lateral columns. Across ten independent arm experiments the classification accuracy at 127 channels (full resolution) resulted in the first and third quartile to be 80-92%. Averaging into four RF signals (simulating 10-mm wide ultrasound beams from each channel) could achieve a median classification accuracy of 87% using the proposed feature extraction method with the discrete wavelet transform. Our results show low resolutions could achieve high accuracies to that of full resolution. We also conducted a preliminary study using a multichannel single-element ultrasound system with lightweight, flexible, and wearable ultrasonic sensors (WUSs) using non-focused ultrasound. Each WUS had an ultrasound sensing area of 20mm by 20mm. Three WUSs were attached on one subject's forearm and ultrasound RF signals were acquired during individual finger flexions. A mean classification accuracy of 98% was obtained with F1 scores ranging between 95 - 98% (across five finger flexions).
Highlights
Trauma related amputation significantly impacts the physical and mental aspects of daily life for upper limb amputees
It is observed that the F1 scores between the discrete wavelet transform (DWT)-mean absolution value (MAV) and ENV-linear regression (LR) methods are similar where the differences are less than 10% for all finger flexions except at the resolution N = 8 where the DWT-MAV method is higher by 14.33% for correctly classifying for middle finger
In this paper, we artificially reduced the lateral spatial resolution of ultrasound RF signals acquired by a clinical linear array ultrasound probe, to investigate the effect of the spatial resolutions on classification performance of finger flexions
Summary
Trauma related amputation significantly impacts the physical and mental aspects of daily life for upper limb amputees. Individuals with an upper limb amputation experience most difficulty with employment, family life, and leisure activities [2]. For industrial upper limb prosthetics, typical non-invasive sensors use surface electromyography (sEMG) to monitor the associated muscle contractions to control the prosthetic robot hand. The commercial prosthetic hand by Touch Bionics ‘‘i-digits’’ uses sEMG to sense general muscle. Commercial sEMG based upper limb prosthetics use around 1-2 sEMG signals as degrees of freedom to sense bulk muscle contractions to directly drive and carry out the select set of discrete motions. The inability of sEMG to detect muscle innervations occurring between different depth regions may limits some applications. Ultrasound has been investigated as an alternative sensing strategy to spatially resolve anatomical muscle activity occurring for different depth regions. Huang et al compared sEMG with B-mode ultrasound imaging by classifying
Sign-in/Register to view highlights & summary Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callDisclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
