Original ContributionsSerial Precision of Metabolite Peak Area Ratios and Water Referenced Metabolite Peak Areas in Proton MR Spectroscopy of the Human Brain
Introduction
In vivo proton magnetic resonance spectroscopy (MRS) has important applications to the human brain, both in the clinical setting and for research purposes including epilepsy,1, 2, 3, 4multiple sclerosis,5, 6, 7brain tumours,8, 9, 10, 11dementias,12, 13, 14, 15stroke,16, 17, 18, 19and other diseases of the central nervous system.20, 21, 22, 23This list of applications and publications is illustrative rather than exhaustive, however.
Clinical magnetic resonance imaging (MRI) is still primarily a qualitative modality where images are visually interpreted by an experienced reader. Although spectra can also be read in such a way, traditionally metabolite peak area ratios have been used as a means of providing a rudimentary quantitation of spectra. The most common metabolite measurements have involved referencing to a measured resonance from the same voxel. In the event of a change or difference in ratios it is not possible to determine which of the two metabolites is responsible for the change, however.20, 27An alternative is to use a reference resonance from a contralateral voxel, or voxel from a similar area of tissue. Again, there are also some disadvantages to such an approach since the contralateral voxel may also be affected despite the presumed unilateral nature of the disease.[16]Clearly even this latter approach is not possible in the case of global metabolic defects. Absolute quantitation avoids some of the limitations associated with metabolite peak area measurements and is made possible by using either a reference standard (which is internal28, 29, 30, 31, 32or external33, 34, 35, 36, 37to the body) or the amplitude of the radiofrequency (RF) pulse required for water suppression.38, 21, 39
Successful MRS depends upon a wide variety of different factors which include patient preparation, voxel shimming, water suppression and choice of acquisition parameters and data post processing to obtain quantitative results. Each of these factors can contribute to the precision of the results obtained. High precision is especially critical for detecting temporal changes due to disease progression and/or treatment. In the clinical realm and also for research studies, it is often necessary to compare results obtained from an individual or group with normative data. In both cases, the precision of measurements is critical to help classify an individual, or to determine whether there are any significant differences between different groups of subjects.
An important factor affecting the precision of MR studies is the degree of manual interaction involved. For MRI, parameters such as the centre frequency and transmit and receive gains are normally determined by an automatic pre-scan procedure. For MRS, however, manual adjustment of a series of parameters such as linear shim offsets and water suppression tip angles is typically required for each acquisition. This manual interaction leads not only to longer scan times, but also means that great care must be taken to achieve high reproducibility using highly trained staff. Once spectra have been successfully acquired, post-processing for quantification may also require interaction, which often can be time consuming and operator dependent and may also therefore add a further random error to the results. Marshall et al. have presented a limited study of precision using a largely interactive approach,[24]some aspects of which have been recently questioned.25, 26They found coefficients of variation (CVs) for “between days” measurements of 10.1–22.6%. In the study reported here, a highly automated spectroscopy acquisition and post-processing protocol was used to avoid the commonly encountered problems described above. We show that even for measurements carried out over a period of 3 months, substantial improvements can be made, with CVs from 2.5–7.7% found.
The work reported here has comprehensively determined the short-term and long-term precision of cerebral in vivo and in vitro metabolite peak area ratios (NAA/Cho, NAA/Cr + PCr, Cho/Cr + PCr) and water referenced metabolite peak areas (NAA/H2O, Cr + PCr/H2O, Cho/H2O) from single voxel proton point resolved spectroscopy (PRESS) MRS (repetition time [TR] = 2000 ms, echo time [TE] = 136 ms, voxel size = 20 × 20 × 20 mm3) over a period of up to 2 years using largely automated shimming, water suppression and post-processing techniques. Automated data analysis consisted of phase-correcting metabolite spectra using information from water reference spectra, zero-filling and Fourier transforming the data, followed by determining the areas of the unsuppressed water and NAA, Cr + PCr and Cho peaks by Levenberg-Marquardt fits to Lorentzian lineshapes. We studied short-term in vitro precision by scanning a phantom five times in a single session on five different occasions. The same phantom was scanned weekly over a period of 2 years to determine long-term in vitro precision. Short-term in vivo precision was determined using five repeat scans on seven subjects in a single session, and long-term in vivo precision determined by five repeat scans on seven subjects over a period of 3 months. Both metabolite peak area ratios and water referenced metabolite peak areas were determined for each of the four studies, and precision characterised in each case by coefficients of variation. Some of the work reported here has previously been presented in a preliminary form.[41]
Section snippets
Theory
Water referenced metabolite peak areas (metabolite peak area/water peak area) may be used as the basis for spectroscopic quantitation.28, 29, 30, 31, 32The concentration of a metabolite M, CM, can be related to the signal from the fully relaxed metabolite (the fully relaxed peak area), SM, and the corresponding concentration and signal for water (CH2O, SH2O). For the three principal metabolites considered here (NAA, Cr + PCr and Cho) this yields the three relationships;
Phantom and Subjects
A phantom containing in vivo levels of metabolites at concentrations found in the adult human brain was used to study in vitro precision. This “liquid brain” phantom (General Electric Medical Systems, Milwaukee, WI, USA) contains 12.50 mM NAA, 10.0 mM Cr + PCr, 3.0 mM Cho, 7.5 mM myoInositol, 12.5 mM glutamate, 12.5 mM lactate, 50.0 mM KH2PO4, 0.1% sodium azide and 0.1% GdDTPA, although only the NAA, Cr + PCr and Cho peaks at 2.02, 3.03 and 3.22 ppm are considered here. The metabolite
Results
The results of each study are presented here in terms of a single or mean (standard deviation) coefficient of variation. To illustrate the similarity of spectra, a figure for each study consisting of five spectra from a single subject, or the phantom, is presented. Because the system gain and centre frequency can vary slightly between studies, each spectrum was scaled to the NAA peak and baseline heights, and the NAA peaks aligned for presentation purposes.
Discussion and Conclusions
Marshall et al.[24]have carried out a study on the reproducibility of cerebral metabolite peak areas (corrected for coil loading and sensitivity) and metabolite peak area ratios obtained from a Siemens 63SP 1.5 T Magnetom MR system (Siemens AG, D-91050 Erlangen, Germany). A PRESS pulse sequence was used (TR = 1600 ms, TE = 135 ms, 256 data averages, 4096 complex data points) and data analysed using Siemens software (NUMARIS) which required manual interaction, and the authors’ own custom
Acknowledgements
We thank the radiographers of the Maudsley Hospital, particularly Caroline Andrews and Amanda Glover for their help in this study, and for suggestions from the two anonymous referees.
References (51)
- et al.
Proton magnetic resonance spectroscopic images and MRI volumetric studies for lateralisation of temporal lobe epilepsy
Magn. Reson. Imaging
(1995) N-acetylaspartate and epilepsy
Magn. Reson. Imaging
(1995)- et al.
Proton magnetic resonance spectroscopy of an acute and chronic lesion in multiple sclerosis
Lancet
(1991) - et al.
Proton magnetic resonance spectroscopy with dementia
Surg. Neurol.
(1993) - et al.
In vivo assessment of N-acetylaspartate in brain in spongy degeneration (Canavan’s disease) by proton spectroscopy
Lancet
(1990) - et al.
Diagnosis of hepatic encephalopathy by proton magnetic resonance spectroscopy
Lancet
(1990) - et al.
Reproducibility of metabolite peak areas in 1H MRS of brain
Magn. Reson. Imaging.
(1996) - et al.
Letter to the editor
Magn. Reson. Imaging.
(1997) - et al.
Absolute molar concentrations by NMR in inhomogeneous B1. A scheme for analysis of in vivo metabolites
J. Magn. Res.
(1983) - et al.
In vivo quantification of brain metabolites by 1H-MRS using water as an internal standard
Magn. Reson. Imaging.
(1993)
The concentration of N-acetyl asparate, creatine + phosphocreatine and choline in different parts of the brain in adulthood and senium
Magn. Reson. Imaging
Regional in vivo proton magnetic resonance spectroscopy of brain
J. Magn. Reson.
In vivo proton magnetic resonance spectroscopy studies of human brain
Magn. Reson. Imaging
Absolute quantitation of water and metabolites in the human brain. I. Compartments and water
J. Magn. Reson. Series B
Absolute quantitation of water and metabolites in the human brain. II. Metabolite concentrations
J. Magn. Reson. Series B
Improved solvent suppression in one and two-dimensional NMR spectra by convolution of time-domain data
J. Magn. Reson.
Simulation of MRI cluster plots and application to neurological segmentation
Magn. Reson. Imaging.
Magnetic resonance spectroscopy in temporal lobe epilepsy
Neurology
Temporal lobe epilepsyPresurgical localisation with proton chemical shift imaging
Radiology
Localised in vivo proton spectroscopy in the brain of patients with multiple sclerosis
Magn. Reson. Med.
Proton magnetic resonance spectroscopy of human brain in vivo in the evaluation of multiple sclerosis. Assessment of the load of disease
Magn. Reson. Med.
Localised proton NMR spectroscopy of brain tumours using short-echo time STEAM sequences
J. Comput. Assist. Tomogr.
On the interpretation of proton NMR spectra from brain tumours in vivo and in vitro
NMR Biomed.
Noninvasive differentiation of tumours with use of localised H-1 MR spectroscopy in vivo. Initial experience in patients with cerebral tumours
Radiology
Human brain tumours. Spectral patterns detected with localised H-1 MR spectroscopy
Radiology
Cited by (59)
<sup>1</sup>H-MRS metabolites in adults with Down syndrome: Effects of dementia
2016, NeuroImage: ClinicalCitation Excerpt :The rationale for selecting this brain region was two-fold (Mann and Esiri, 1989): to reduce the impact of movement artifacts, which can be a concern with imaging people with DS and (Wisniewski et al., 1985); to select a brain region that is sensitive to mild cognitive impairment and AD in the general population in previous MRS studies (Lin et al., 2005; Tumati et al., 2013). A Stimulated Echo Acquisition Mode (STEAM) sequence was used with repetition time (TR) of 1500 ms and echo time (TE) of 35 ms, flip angle = 90°, 128 averages and 1024 points; automated local shimming and water suppression (Simmons et al., 1998). The rationale for using STEAM was to allow us to compare our current results with ongoing studies of MRS in people without DS with mild cognitive impairment and AD at our imaging center and previous publications (Tumati et al., 2013; Murata et al., 1993).
Design of a fused phantom for quantitative evaluation of brain metabolites and enhanced quality assurance testing for magnetic resonance imaging and spectroscopy
2015, Journal of Neuroscience MethodsCitation Excerpt :However, the possibility of QA testing was demonstrated by our results using our designed phantom with optimized parameters (Table 1). Various MRI and MRS phantoms have been developed for metabolite quantification (Rice et al., 1998; Simmons et al., 1998), relaxation times of human brain tissue (Rice et al., 1998), angiography (Krämer et al., 2004), MRI human tissue equivalent (Hattori et al., 2013; Kato et al., 2005; Ikemoto et al., 2011), reproducibility of geometric distortion in MRI (Mizowaki et al., 2000), high-resolution phantom for MRI (Fellner et al., 2001), and MRS QA phantom (Woo et al., 2009). MR phantoms do not obtain information regarding both images and spectra in the MR scanner in the same session.
Proton MRS and MRSI of the brain without water suppression
2015, Progress in Nuclear Magnetic Resonance SpectroscopyDown syndrome with and without dementia: An in vivo proton Magnetic Resonance Spectroscopy study with implications for Alzheimer's disease
2011, NeuroImageCitation Excerpt :The center of the voxel was denoted by the center of the white matter tract in the superior/inferior and right/left positions. A point-resolved spectroscopy (PRESS) pulse sequence (TE 35 ms, TR 2000 ms, 160 data averages, 2048 points) with automated local shimming and water suppression and excellent reproducibility (Simmons et al., 1998) was used to obtain spectra from each voxel with high signal to noise ratio and clearly resolved NAA, mI, Cr + PCr and Cho peaks. Water suppression was carried out by a sequence of chemical shift selective (CHESS) radio frequency (RF) pulses to excite and associated spoiling gradients to dephase water before the localization sequence.
Limits of a localized magnetic resonance spectroscopy assay for ex vivo myocardial triacylglycerol
2007, Journal of Pharmaceutical and Biomedical AnalysisDevelopment of a cone-shape phantom for multi-voxel MR spectroscopy
2007, Journal of Neuroscience Methods