Quantitative in vivo magnetic resonance spectroscopy using synthetic signal injection

Abstract

Accurate conversion of magnetic resonance spectra to quantitative units of concentration generally requires compensation for differences in coil loading conditions, the gains of the various receiver amplifiers, and rescaling that occurs during post-processing manipulations. This can be efficiently achieved by injecting a precalibrated, artificial reference signal, or pseudo-signal into the data. We have previously demonstrated, using in vitro measurements, that robust pseudo-signal injection can be accomplished using a second coil, called the injector coil, properly designed and oriented so that it couples inductively with the receive coil used to acquire the data. In this work, we acquired nonlocalized phosphorous magnetic resonance spectroscopy measurements from resting human tibialis anterior muscles and used pseudo-signal injection to calculate the Pi, PCr, and ATP concentrations. We compared these results to parallel estimates of concentrations obtained using the more established phantom replacement method. Our results demonstrate that pseudo-signal injection using inductive coupling provides a robust calibration factor that is immune to coil loading conditions and suitable for use in human measurements. Having benefits in terms of ease of use and quantitative accuracy, this method is feasible for clinical use. The protocol we describe could be readily translated for use in patients with mitochondrial disease, where sensitive assessment of metabolite content could improve diagnosis and treatment.

Figure 1. The key hardware components required to inject the pseudo-signal.

Prior to execution of the pulse sequence, a Unix macro was used to create a digitized waveform describing the desired pseudo-signal. The pulse sequence read the waveform and sent it to the second RF channel (RF2). The pseudo-signal passed through an external attenuator (Ext atten) before being fed to the injector coil. During sequence execution, the surface coil was operated in transmit/receive mode while the injector coil was used only to transmit the pseudo-signal during the acquisition window (AQ).

Figure 2. A sample in vivo spectrum with real peaks and the artificially injected pseudo-peak.

The Pi, PCr and the γ, α, and β moieties of ATP appear at about 2.5, 0.5, −2.5, −5.0, −10, and −18.5 ppm, respectively. The pseudo-peak appears at 41 ppm. The amplitude, frequency, phase, and line-width of the pseudo-peak are determined by the digitized waveform transmitted by the RF synthesizer and are easily adjusted.

Figure 3. Concentrations determined by the pFID method are accurate and independent of coil loading conditions.

The nominal concentrations (x-axes) were determined using the phantom replacement method and calibration measurements obtained from a phantom that replicated in vivo coil loading conditions. The nominal concentrations reflect the normal ranges of Pi, PCr and ATP expected in healthy male subjects. All three sets of pFID measurements fall very close to the lines of identity (dashed lines), validating that this method allows accurate quantification of metabolite content for a wide range of coil loading conditions. In contrast, the concentrations determined using the phantom replacement method were invalid when in vivo coil loading conditions were not replicated in the calibration session. When the calibration phantom overloaded the coil, the phantom replacement method overestimated the in vivo concentrations. When the phantom underloaded the coil, the phantom replacement method underestimated the in vivo concentrations.

Figure 4. Typical metabolite concentrations from one subject and the mean (+/− standard deviation) metabolite concentrations from all subjects.

The pFID method provided accurate estimates of metabolite content in each subject that were in agreement with the nominal concentrations. The pattern was consistent for all subjects, as reflected by the bar graph of the mean concentrations. The standard deviations primarily reflect the true range of concentrations within the subjects (as seen in Figure 3) rather than random variations in the measurements.