The Hisense* medical equipment company is developing a new ultrasound platform with related ultrasound desktop and portable devices.

The platform adopts a new GPU and CPU heterogeneous computing architecture. To provide developers with a more flexible experience, the platform places running more post-processing functions on the CPU side, which contains some functional modules with more complex operations.

The modules, compiled with traditional CPU coding, have low efficiency and high time consumption. Based on this, the new platform uses Intel oneAPI solutions to adopt different acceleration schemes for module characteristics. For modules with high complexity and long calculation times, a SYCL* schema based on Intel® HD Graphics is adopted for coding implementation.

Through comparison and verification, the total running time of post-processing has been reduced, which meets the real-time processing requirements of the ultrasonic system for signals and images.

Speakers

Chuanqing Fu is engaged in the system research and development of ultrasound equipment. He is a senior algorithm engineer who is mainly responsible for designing the image post-processing scheme of the ultrasound system and optimizing ultrasound images.

Zhenzhong Liu works in developing software for ultrasound equipment. He is a senior software engineer who is responsible for designing the ultrasound software system architecture and designing the signal and image-processing software acceleration scheme.