Each advancement in medical imaging enables significant improvements in the quality and efficiency of diagnostics and therapy. The challenge: Each advancement requires significant computing performance gains to allow medical image data to be measured and analyzed. And once this massive amount of digital data has been created, it needs to be visualized too – including even stereoscopic 3D in real-time. With CUDA from NVIDIA, OpenCL from Khronos Group and Xeon Phi™ from Intel, several alternative massive parallel processing technologies are available for these tasks – all with their own benefits and restrictions. So which technology is right for data processing on medical backend systems? And how can this technology be combined most efficiently with different processor boards, operating systems and graphic algorithms? This whitepaper helps engineers understand the major differences in technology and identify the most appropriate framework for building their professional, medical-grade image processing systems.