Twieg D B
Magn Reson Med. 1985 Oct;2(5):437-52. doi: 10.1002/mrm.1910020504.
The echo-planar imaging method (EPI) and its variants can acquire data sufficient to form an image in a much shorter time than the Fourier or projection-reconstruction methods. Fundamental requirements on gradient waveforms and signal sampling are discussed for EPI and SEPI (a variant using sinusoidal gradient waveforms), based on the k-trajectory formulation of acquisition in NMR imaging. Using a new simulation method, image defects in EPI and SEPI are demonstrated which result from failure to meet these requirements. Faithful rendition of theoretically ideal gradient waveforms places severe demands on gradient system response in rapid imaging. It is shown, however, that nonideal gradient system response need not degrade images if appropriate reconstruction computations are used. Transfer functions and noise power spectra are derived for EPI and SEPI, and the ratio of these two functions is used to derive ideal observer signal-to-noise ratios for simple visual interpretive tasks. It is shown that ideal observer performance can differ significantly between EPI and SEPI for some visual tasks, using the same NMR imaging system.
回波平面成像方法(EPI)及其变体能够比傅里叶或投影重建方法在短得多的时间内采集到足以形成一幅图像的数据。基于核磁共振成像中采集的k轨迹公式,讨论了EPI和SEPI(一种使用正弦梯度波形的变体)对梯度波形和信号采样的基本要求。使用一种新的模拟方法,展示了EPI和SEPI中由于未满足这些要求而导致的图像缺陷。在快速成像中,理论上理想梯度波形的忠实再现对梯度系统响应提出了严格要求。然而,研究表明,如果使用适当的重建计算,非理想的梯度系统响应不一定会降低图像质量。推导了EPI和SEPI的传递函数和噪声功率谱,并使用这两个函数的比值来推导用于简单视觉解释任务的理想观察者信噪比。结果表明,对于某些视觉任务,使用相同的核磁共振成像系统时,EPI和SEPI之间的理想观察者性能可能会有显著差异。