Computational protocol: In vivoH1 MR spectroscopy using 3 Tesla to investigate the metabolic profiles of joint fluids in different types of knee diseases

Similar protocols

Protocol publication

[…] Spectroscopy data were saved in a Philips format and transferred into a personal computer. The data were analyzed using public freeware software called jMRUI (Academy of Sciences of the Czech Republic, Brno, Czech Republic, http://www.jmrui.eu/support/scientific-technical-literature/). After the chemical shift of the water was set to 4.7 ppm, spectroscopic data were created by numeric integration over a range from 0 to 8 ppm. Spectra were processed with zero‐order phase correction based on the water peaks using Lorentzian apodization (3 Hz). Because water signals are not perfectly suppressed by CHESS, a Hankel‐Lanczos singular value decomposition (HLSVD) filter was applied to the spectrum‐suppressed water signals in postprocessing to subtract the residual water signals in the frequency domains. A nonlinear least square algorithm, called AMARES in the jMRUI software package, was used to fit the spectrum, using the Gaussian line shape for each metabolite's resonance and a Lorentzian line shape for the water peaks. Resonance peaks were assigned based on the analysis of both characteristic and previously published chemical shift values., , , , The identities of the components responsible for the resonances in the proton MR spectra of the joint fluid samples were assigned by considering their characteristic chemical shift values. In this study, the metabolites of CH3 (0.7~1.1 ppm), CH2 (1.1~1.5 ppm), and CH=CH (5.1~5.5ppm) lipids as well as water were obtained from all patients. We grouped both CH3 and CH2 lipids in the low ppm range and CH=CH lipid in the high ppm range because the CH3 and CH2 lipids usually overlapped. If the CH2 and CH3 metabolites in the low ppm range were dominant, then we designated this spectrum Type A. If the CH=CH metabolite in the high ppm range was dominant, then the spectrum was Type B. Finally, if the metabolites in both the low and high ppm ranges were dominant, then the spectrum was Type C. In order to quantify the three types of spectra, we calculated the lateralization index (LI) as(1)LI=(PA−PB)(PA+PB)where PA was the sum of the peak intensity of CH2 and CH3 and PB was that of CH=CH. Therefore, LI values should have been between 1.0 and −1.0. LI=1.0 meant that the CH2 and CH3 lipids were dominant in the spectrum (Type A spectrum); LI=−1.0 meant that the CH=CH lipid was dominant (Type B spectrum); and LI=0.0 meant that both Type A and B metabolites were balanced in the spectrum (Type C spectrum). [...] Before statistical evaluations of metabolites, normality was tested using the Kolmogorov‐Smirnov test for each metabolite. None of the metabolites had a normal distribution, and therefore, we used a nonparametric test for the following investigations. First, we used the Kruskal‐Wallis test to investigate any differences in metabolites among the three patient groups (degenerative, traumatic, and infectious and inflammatory diseases) and then the Mann‐Whitney test to compare metabolites between the degenerative and traumatic diseases. Second, to investigate the relationships between metabolites, we calculated the Spearman's coefficient of rank correlations (rho) from the CH2,CH3, and CH=CH lipids and the sum of the CH2 and CH3 lipids. Finally, receiver operating characteristic (ROC) curve analysis was performed to investigate sensitivity and specificity using metabolites in the degenerative and traumatic disease groups. Statistical analysis was performed with MedCalc Statistical software Version 15.4 (MedCalc Software bvba, Ostend, Belgium; http//www.medcalc.org; 2015) and with significance set at p=0.05. […]

Pipeline specifications

Software tools jMRUI, MedCalc
Applications Miscellaneous, NMR-based metabolomics
Organisms Homo sapiens