· Settings for RF transmit gain (TG), transmit/receive frequency (center frequency or CF) and receiver gain (RG) are patient-dependent and are critical to good image quality.

· Finding optimal settings for these parameters is similar to the problem of mapping the parameters but over a smaller volume, usually one slice near the magnet isocenter.

· The parameters must be set robustly, quickly and without operator intervention.

· Different vendors use different methods for measuring and calibrating these parameters.

· Non-optimal RF transmit gain can result in loss of signal and contrast, poor image uniformity, and poor fat saturation.

· Non-optimal transmit/receive frequency can result in banding and artifacts in SSFP, blurring with spirals, phase encode shifting in EPI, poor image intensity homogeneity, localization errors, signal loss, and poor fat saturation.

· Non-optimal receive gain can cause shading, intensity distortion, loss of SNR and artifacts.

Scientists and clinicians interested in patient-dependent scanner calibration methods and the effects on image quality.

· Understand ways of setting the RF transmit/receive gain and transmit/receive frequency.

· Understand possible artifacts that result if the values are not set reasonably well.

· Understand situations in which the calibrations could be prone to being incorrectly set.

Scanners are calibrated using phantoms for patient-independent effects such as large scale B0 inhomogeneity, eddy currents, RF output power, gradient amplifier/coil gain and gradient non-linearity. Once these calibrations are completed the scanner is ready for clinical use. The RF transmit power, RF receiver attenuation and transmit/receive frequency must be set for each patient to obtain optimal image quality. Various methods are available to set the parameters but in general the methods must be fast and robust to patient and protocol variation. A variety of artifacts can also arise depending on how the parameters are set.

Differences in patient size and composition within the sensitive volume of the transmit coil change RF energy absorption. As a result the RF power needed to obtain the desired flip angle must be calibrated for each patient at the desired scanning location. The same methods that are used for B1 mapping can be used for this purpose. Examples of such methods include the double angle method (DAM) (1), actual flip angle imaging (AFI) (2), phase sensitive imaging (3), dual refocusing echo acquisition mode (DREAM) (4), and Bloch-Siegert (5). However for TG estimation, only a small subset of the patient imaging volume is mapped, usually within a slice at the magnet isocenter. The TG calibration sequence is used to find the flip angle that results from a given TG for some standard RF pulse. The TG needed for the flip angle for any other pulse can then be obtained by scaling. The process need not be repeated unless patient loading is substantially changed.

The attenuation of the analog signal or equivalently the receive gain (RG) must be adjusted prior to digitization by the A/D converter. If the RG is too high the signal will saturate the A/D converter resulting in shading and intensity distortion. If the RG is too low, the noise will be dominated by the step size of the A/D converter instead of the noise itself (quantization error), resulting in noise and artifacts (6,7). Since the highest signal occurs at the center of k-space, the imaging pulse sequence can be used to acquire all or part of the imaging volume without phase encoding. If the system is well designed, the RG can be adjusted so that the signal maximum is somewhat less than the amplitude that saturates the A/D while still avoiding quantization error. The system design requirements depend on the dynamic range of the analog signal (8). The highest dynamic range results from the highest SNR scans. These are usually high field, 3D scans with receive coils that are sensitive to a relatively large patient volume, as well as scan protocols that generate high signal.

The transmit/receive frequency or center frequency (CF) must also be adjusted for each patient and sometimes several times during the exam depending on the scanned location. A large CF error can result in localization errors in the slice direction. With EPI, because of the low effective bandwidth in the phase encoding direction, a large CF error can shift the object in the phase encoding direction. With balanced SSFP, CF errors can cause banding artifacts (9). With spirals, CF errors cause blurring artifacts (10). With some fat saturation methods, a CF shift can result in poor fat saturation and loss of water signal.

Finding the optimal TG for the diagnostically relevant anatomy can be difficult if the transmit B1 varies strongly over the patient, for example at high field where there is strong dielectric shading or when the anatomical shape is complicated. In these cases a compromise must be made between image quality and having a fast, automated way to set TG.

When adjusting the RG, some receive coils can have a large enough phase variation over the imaging volume that the maximum signal does not occur at zero phase encoding. In this case, a phase encoding value near zero can partially cancel the receive coil phase and produce the maximum signal. Therefore it is prudent to set the RG to produce somewhat less than the maximum signal that results when zero phase encoding is used.

The CF can be particularly difficult to set in some areas of the body with poor B0 homogeneity such as the neck due to rapid local susceptibility variations. Therefore it is advantageous to set CF over a small localized volume and after or in conjunction with some sort of local shimming procedure. CF can also be difficult to set in areas that have multiple, complicated spectral peaks, or when the main fat peak is actually larger than the water peak, for example in breast imaging. Therefore a robust fat elimination method can be helpful in finding the water peak.

RF transmit gain, receive gain and transmit/receive frequency must be adjusted for optimal image quality for each patient exam and sometimes further adjusted within the patient exam. Calibration must be robust, fast and automated. Different vendors use different methods. Non-optimal settings can result in artifacts, poor contrast, SNR loss, and poor fat suppression.