Shape Sensing for Wearable Ultrasound

Abstract

Wearable ultrasound technology has the potential to enable continuous, non-invasive monitoring of physiological processes, offering new opportunities for diagnostics and patient care. To achieve this, ultrasound transducer arrays must be flexible enough to conform to the human body. However, deformation of a flexible array changes the positions of individual transducer elements, which can significantly degrade the accuracy of ultrasound localisation and imaging algorithms. This thesis investigates the feasibility of using dedicated shape sensors to compensate for these deformations and thereby improve localisation accuracy in wearable ultrasound systems. A concept demonstrator consisting of two rigid ultrasound transducer arrays connected by a flexible substrate was developed. A shape estimation pipeline was developed to reconstruct the geometry of the substrate from sensor measurements and a localisation algorithm was implemented to estimate the position of a point reflector using ultrasound measurements. Both methods were validated individually in simulation and subsequently evaluated together in both simulation and a hardware demonstrator. The results demonstrate that accurate shape estimation and reflector localisation can be achieved using compact strain-based sensing.
Furthermore, the integrated system was able to align localisation estimates from both transducer arrays with sub-millimetre agreement, demonstrating successful compensation for array deformation. It is therefore concluded that shape-sensor-based deformation correction is a promising approach for wearable ultrasound systems. Future work should focus on increasing the spatial density of the shape sensors, improving shape interpolation and reconstruction algorithms, and extending the method to two-dimensional array geometries to support practical imaging applications.