A University of South Wales PhD student has developed a new algorithm which has the potential to change the way patients with severe neuromusculoskeletal conditions are assessed and treated.
Working in collaboration with the Cardiff and Vale University Health Board, Sebastian Haigh is currently in his second year of his KESS 2 funded scholarship, entitled ‘High Accuracy System for Anatomical Landmarks Localisation using Ultrasound’.
His work could help those with limited body movement by investigating the feasibility of automatic measurement and recording of anatomical landmarks, representing the skeletal configuration of clients of clinical services, such as the Posture and Mobility Service.
Published in IEEE TRANSACTIONS journal, Sebastian’s paper develops a novel algorithm for the classification and rejection of non-line of sight reflections in narrowband ultrasonic localisation systems.
Non line of sight reflections occur when the transmitted ultrasonic waves bounce off solid objects and travel through any path that does not follow the direct line of sight between transmitter and receiver.
Sebastian explained, “Reflections result in a situation where multiple signals can be received from a single ultrasonic transmission, only one of which has the possibility of having travelled the required line of sight path. Determining which signals are non-line of sight reflections is challenging in narrowband systems and failing to do so can result in a loss of accuracy.
“Our algorithm uses a Bayesian probabilistic iteratively reweighted least squares approach to detect and exclude reflected signals by adjusting a set of weights based on several factors, including how large an error signals generate in calculating the position. This method can classify over 96% of signals correctly while also outperforming state of the art algorithms in terms of computational complexity.”
It is estimated that there are approximately 20,000 people across the UK currently, who require custom contoured seats to meet their clinical and functional needs and could benefit from this technology. If successful this new technique will enable a non-invasive, cost effective measurement of anatomical landmarks and the objective assessment of posture over time.
The published IEEE journal paper can be found at : https://ieeexplore.ieee.org/document/8428614/
Further details and background info can also be found on the project group webpage : https://at-web1.comp.glam.ac.uk/KBS/alandamarks.php
