Fast ultrasound signal and image processing on a tablet device

During the last decade, portable ultrasound devices have been proven beneficial in a variety of situations (e.g. OB/GYN, point of care and cardiovascular applications). The aim of this work is to investigate the feasibility of implementing standard signal and image processing algorithms on a tablet device, instead of the scanner´s backend unit. Furthermore, the additional benefit of using the tablet´s graphical processing unit to offload the CPU intensive tasks and balance the battery consumption was tested. A plugin for streaming data over a wired or wireless TCP/IP connection from an ultrasound scanner was implemented and data sent to a tablet device which runs the image and signal processing algorithms. The use of compute shaders for further optimizations was investigated. The prototype was tested by midwives, both experienced and inexperienced in the use of ultrasound, in Norway and South Africa. Afterwards they were asked to fill out standard system usability scale forms. Conclusively, most of the backend components of an ultrasound scanner can be implemented on a tablet´s GPU. Compute shaders have a significant impact on battery consumption, which can be reduced by a factor of up to 3 and as such significantly extend the scanning time of a portable system.