Supervisors: Dr. Sergio Sanabria, Dieter Schweizer, Prof. Dr. Orcun Goksel
Conventional ultrasound (US) B-mode systems can be utilized to detect tumors and other irregularities in human tissue. However, they do not provide quantitative diagnostic information to differentiate benign and malignant lesions. The Computer-assisted Applications in Medicine (CAiM) group has developed a hand-held speed-of-sound (SoS) imaging method, which allows quantification of tumorous lesions. Even if this method is in principle supported by conventional ultrasound hardware, it requires low-level access to the ultrasound signals recorded by the probe, which is typically not available in the data acquisition pipelines of clinical ultrasound systems. In particular, full-matrix data acquisition is required, which means that all receive channels must be recorded for each individual transmit firing. This will allow implementation of sophisticated computed tomography, motion tracking and synthetic aperture algorithms. Until now, it was only possible to obtain this data from a research US system. These systems are either not approved for clinical use or show an inefficient workflow that makes this acquisition in a clinical setting challenging. Due to usability reasons, such as having a large time overhead for switching between B-Mode imaging and SoS imaging, this was not suitable for diagnoses in a clinical environment. This semester project was done in industry-research collaboration with a Swiss ultrasound manufacturer, Fukuda Denshi Switzerland AG, and aimed at implementing full-matrix data acquisition in a clinical ultrasound system UF-760 AG. The UF-760AG's beamformer FPGA was adapted such that raw acquired US data samples can be transported in real time to a remote PC system. As a second step, the UF-760AG was extended with a novel so-called SoS Mode, in which a specific sequence of capturing B-Mode frames and SoS raw data frames is executed, resulting in a full-matrix raw data transfer of 188 MB at 1.9 Hz (360 MB/s) to the internal memory of the PC. Full-matrix raw data is transmitted with time stamps inline with B-mode, which allows both identification and quantification of breast lesions. A cluster of 4 1TB SSD allows for 2 hours of clinical data recording. A control channel allows configuration of the ultrasound system from the PC. This novel system meets the hardware requirements for constructing SoS images in line in parallel with B-Mode images and facilitates realistic and conventient clinical examinations.