An alternative, but equivalent way of understanding why SNR depends on RBW is to consider the frequency-domain behavior. Because thermal noise is spread uniformly across the whole radiofrequency band, decreasing the RBW means that less noise is collected along with the MR signal. For a given FOV, the smallest possible RBW will yield the highest SNR. For a given FOV, it is possible to acquire an image using a range of RBWs. As illustrated in Figure 1.4, the combination of the RBW and FOV together determine what the readout gradient amplitude (Gx) must be to map this range of frequencies to the specified FOV. On a Signa scanner, Gx is changed automatically by the system, with the user selecting both the FOV and RBW.

Figure 1.4 The RBW and FOV are user-selected on the Signa scanner, and the gradient strength in the readout directin, Gx, is adjusted appropriately. For larter RBW, the gradient strength must be increased to keep the FOV the same.

There is a limit to how small a RBW can be selected for a particular choice of timing parameters. Decreasing the RBW requires that the signal must be sampled more slowly. Thus, for the same number of acquired data points, Xres, the acquisition window is necessarily longer. In a spin echo, or gradient echo pulse sequence, the length of the acquisition window is limited by the userÕs choice of TE. If the acquisition window is too long to fit in the specified TE, an error will be generated. In a fast spin echo (FSE) pulse sequence, the choice of RBW is flexible (when Variable Bandwidth is selected), and is specified by the user. (See Fig. 1.5) However, the effective TE for FSE (TEeff ) must be adjusted by the scanner from the user-selected value, based on the minimum echo spacing (ESP) that can be achieved for the chosen RBW. TEeff is defined as the echo that is used to fill the center of k-space in an FSE acquisition, and must be an integral multiple of the ESP. Shorter ESP (obtained using higher RBW) therefore means that the user can more precisely select the TEeff for the sequence. (For a more detailed discussion of the FSE sequence, see Section 1.4.3).

Figure 1.5 The RBW specifies the rate at which data is sampled in the acquisition window. For higher RBW, the data is sampled more rapidly, and the acquisition window for a given Xres is shorter.

While a lower RBW is preferred for higher SNR, chemical shift misregistration and susceptibility-induced distortion artifacts are more severe. A lower RBW for a constant FOV results in lower bandwidth per pixel, i.e. each pixel location is encoded with a narrower range of frequencies. This means that MR signals at frequencies different from the central frequency of water will be mapped to pixels farther away from their actual location, resulting in more pronounced susceptibility and chemical shift artifacts. As will be discussed in more detail in Chapter 10, post-surgical patients who have metal instrumentation should be imaged at higher RBW to reduce the imaging distortions resulting from susceptibility gradients created by the magnetized metal components. For these patients, reducing image distortions from metal artifact is a much more important consideration than maximizing the image SNR.