基于油水分离的核磁共振并行温度成像及噪声分析

基于油水分离的核磁共振温度成像能够有效减少常规水质子共振频率测温法的各种误差,但成像速度受到多梯度回波序列TR时间的限制。本文使用SENSE并行成像技术对成像过程进行加速,并建立相应的噪声分析模型。同时提出逐通道重建方法解决自标定SENSE技术中的相位缺失问题,并采用CORNOL正则化方法降低高加速倍数时的温度图噪声。定性和定量数值模拟结果验证了并行温度成像方法和噪声分析模型的有效性。体模实验结果表明,若采用正则化SC-SENSE重建技术,使用4通道头部线圈时净加速倍数可达2.7倍。

Water-fat-separation based MR parallel temperature imaging and noise analysis

Water-fat-separation based MR thermometry can effectively reduce the errors of conventional PRFS MR thermometry,but its temporal resolution is limited by the repetition time(TR) of multi-echo GRE sequence.SENSE parallel imaging method is used to accelerate the imaging process and the noise analysis model is established in this paper.Channel-by-channel reconstruction procedure is proposed to deal with the missing phase problem in SC-SENSE,and CORNOL regularization is adopted to reduce the noise level of temperature maps at high acceleration ratio.The effectiveness of the parallel temperature imaging method and noise model was validated by qualitative and quantitative results of numerical simulation.The result of whipping cream phantom experiment indicates that the net acceleration ratio of 2.7 can be achieved with 4-channel head coil by the SC-SENSE reconstruction and CORNOL regularization.