Acceleration of ultrasonic data compression using OpenCL on GPU

Abstract

Translate Article

Ultrasonic imaging applications require real-time acquisition and processing of huge volume of data. Subsequently, reducing the data storage becomes an essential requirement for applications involving ultrasound technology. Compression of the acquired data significantly reduces the storage and also helps in faster data processing. Furthermore, computationally efficient data compression enables real-time data transfer to remote locations for expert analysis. In this study, an efficient implementation of ultrasonic data compression based on discrete wavelet transform (DWT) is realized on GeForce GT 750M NVIDIA graphical processing unit (GPU) using Open Computing Language (OpenCL). Execution of OpenCL program on GPU allows parallel computing of the compression algorithm to accelerate the computational performance. This study demonstrates that OpenCL implementation of ultrasonic compression algorithm on GPU requires only 0.29 seconds for compressing 33 Mbytes of ultrasonic data into 6.6 Mbytes (80% compression), which makes it suitable for real-time ultrasonic imaging applications.