Research Of MRI Real-Time Signal Receiving And Image Reconstruction

Magnetic resonance imaging (MRI) signal receiving is closely related to the quality of reconstruction image. Therefore, it's important to study high-performance signal receiving methods. Because dedicated digital down converter (DDC) is employed as the core device of digital receiver, traditional MRI signal receiving scheme is not flexible enough and the precision is limited, affecting the image signal-to-noise ratio (SNR), which has become an important factor restricting the imaging quality. To solve this problem, high-speed data acquisition card is employed to sample the echo and the algorithms and hardware related to digital receiver are developed.The research on the algorithms and the design of multi-stage filter and decimation was completed. Parks-McClellan design method is employed to design equiripple filters. Through optimizing each filter's passband tolerance, stopband tolerance and transition bandwidth, the order of the filter was effectively reduced, and the computation efficiency was improved as well.In recent years, the graphics processing unit (GPU) with programming flexibility was produced. The processing speed can achieves teraflops per second. A technical solution based GPU directly processing the echo data (including digital mixing, filtering and decimation and Fourier reconstruction) was proposed with the first time in the industry. Thus, it can flexibly achieve the filtering algorithms and the highest calculation accuracy. Then, these equiripple filters were applied to the designed digital receiver with increasing image SNR, and optimization of GPU program was implemented to effectively reduce the time consumption of signal receiving and image reconstruction which achieved the effect of real-time processing.In addition, in view of TI's multicore DSP-TMS320C667x series are a highest-performance floating-point DSP, the research on digital receiving using multi-core DSP and the tests of the time consumption of signal receiving and image reconstruction were completed.Experimental results demonstrate that the designed signal receiver based on the data acquisition card and GPU can significantly reduce the time consumption of imaging and improve image SNR and meet the needs of real-time signal receiving and image reconstruction of MRI.