Introduction

Glioma tumors are differently graded in terms of major malformations in adults [ 1 ]. Diagnosis and grading them with high accuracy in a non-invasive manner is very important to determine a correct treatment plan and, in some cases, to prevent aggressive surgical treatment [ 2 , 3 ]. Grade histological information and tissue type in brain tumors are important for clinical management of patients, which have a close relationship with patients’ survival probability. However, there are two major limitations in the grading and histologic diagnosis of brain tumors, especially in Glioma. Initially, the sampling error through Stereotactic method can be mentioned. In some cases, the sample cannot be considered as representative of the total volume of the tumor. Moreover, it is difficult to correctly evaluate the remaining tumor tissue after cytoreductive surgery [ 4 ]. Contrast-enhanced magnetic resonance imaging (MRI) is currently used as a gold standard method to guide the biopsy by neurosurgeons. However, the results of this technique are sometimes ambiguous [ 5 ]. MRI is recognized as the most common imaging modality in the evaluation of intracranial tumors [ 6 ]. MRI is an excellent method for structural and anatomic diagnosis of the brain, but it does not provide information on vascularity, metabolism, and cellularity that are important for tumor diagnosis and grading [ 7 ]. For this reason, the use of MRI advanced techniques such as MRS is important. Proton magnetic resonance spectroscopy (¹H-MRS) is a noninvasive method which can provide information about neuronal integrity, cell proliferation, cell degradation, and energy metabolism in brain tissue [ 8 ]. In the brain, ¹H-MRS can be used to calculate various metabolites like N-acetyl aspartate (NAA), choline compounds (Cho), creatine and phosphocreatine (Cr), lactate and lipid [ 9 ]. ¹H-MRS can play an important role in making brain tumor diagnosis more accurate, which can represent the metabolic changes before observing the structural change on MRI [ 10 ]. The ¹H-MRS range is expanding at the clinic [ 11 ], but using this method requires a proper post-processing process to properly quantify the spectrum obtained from the test [ 12 ]. Spectra of magnetic resonance spectroscopy are analyzing and quantifying to determine cerebral metabolites using two types of software packages: a java-based version of the MR user interface package (jMRUI v. 5.0) [ 13 ] and the totally automatic robust quantitation in nuclear MR (TARQUIN) algorithm [ 14 ]. JMRUI is a highly flexible software package that provides a wind range of algorithms for ¹H-MRS signal processing, and includes preprocessing tools and peak fitting. TARQUIN is a new algorithm, free to use and change under the general public license, accessible on computer platforms and designed to provide rapid and automatic metabolite quantitation. In a fully automated operation, baseline interference is minimized by point truncation and HSVD water removal [ 15 ], thus eliminating user variability. Point truncation removes the very early points of the free induction decay, which includes very large signals that are difficult to model and the last ones that only contain noise. However, studies have been conducted to assess the sensitivity of quantification approaches in certain diseases [ 16 , 17 ], but no report evaluated the quantification models in the analysis of Glioma tumor metabolites according to available data. The purpose of this research was to compare two models of ¹H-MRS data analysis to identify changes in the metabolite in Glioma tumors.

Material and Methods

In this analytical study, the inclusion criterion was people whose Glioma tumor was confirmed by histologic information. Two post-processing methods of java-based graphics for MR user interface packages (JMRUI) and fully automated robust quantitation in NMR (TARQUIN) were used to quantify the H-MRS results. Data analysis was performed on 15 patients with glial brain tumor and 15 control cases. In this research, the differences in metabolism quantification were investigated using these two models in Glioma tumor. Each subject gave their written informed consent after the procedure was fully explained and understood.

1H-MRS

¹H-MRS imaging was performed on a 1.5-T scanner (Siemens MAGNETOM Verio; Siemens Medical Systems, Erlangen, Germany) at Qaem Hospital, Mashhad, Iran. In order to ensure a precise location of the voxels, T1 weighted brain MR images were taken on the sagittal, coronal and axial planes and T2 weighted images were collected to remove any apparent cerebral pathology. A manual of procedures for MRS spectroscopy (MRS) was used to ensure that all MRS exams were carried out using similar operating settings. Voxels were put on areas of the brain that appeared anatomically abnormal in images T1 and T2 for glioma group and normal areas for control group. Subjects were told to lie down in a supine position. The Circular polarized (CP) head coil was then placed over the head for both image and ¹H-MRS acquisition. First, localization images were obtained, then T2 weighted images were obtained in the coronal, axial and sagittal planes (Echo time (TE)=30 ms, repetition time (TR)=1500 ms, and slice thickness =6 mm) to ensure voxel localization. The automatic shimming protocol available on MRI was performed. Later, single-voxel ¹H-MRS acquisition for metabolites assessment was done with point resolved spectroscopy (PRESS) pulse sequence (TE=43 ms, TR=2000 ms, Number of signal averages =156, Data point =2048, and band width=2500 Hz) with water suppression. Voxel size of 20×20×20 mm³ was carefully placed in glioma tumor. Cerebral metabolites N-acetyl aspartate (NAA), Myo-inositol (Mi), and Choline (Cho) have been measured and expressed as creatin ratios (Cr). All spectra were visually checked by an experienced physicist to ensure acquisition efficiency and the acquisition was repeated in cases with low signal-to-noise ratio.

Analysis models

In this study, an advanced method for accurate, robust, and efficient spectral fitting (AMARES algorithm) was applied for spectra fitting in time domain [ 18 ] in JMRUI software. JMRUI provides two-stage time domain analysis of in vivo MRS data. Pre-processing involves user interaction with the HLSVD / HLSVDPro filters to remove residual water molecules [ 18 ] and the Cadzow function is used to filter the signal [ 12 ]. This manual pre-processing step will impact model fitting results and hence affect the accuracy of the signal quantification. For all analyses in this report, the same prior knowledge of the approximate peaks was input, with peaks set at the following positions [ 19 ]; 2.0 parts per million (ppm) and 3.9 line width for NAA, 3.01 ppm and 4.9 LW for Cr, 3.2 ppm and 4.9 LW for Cho and 3.54 ppm and 4.9 LW for mI. TARQUIN has more advantages than jMRUI since it enables automatic post processing for spectra metabolites, and wide ranges of ¹H-MRS data can be accepted. TARQUIN is an algorithm which suits a time domain by a least square projection used to determine signal amplitude. Notable features are that TARQUIN imposes soft constraints with basis in-vivo spectra data set that includes macromolecules, lipids, and metabolites to avoid possible over fitting of spectrum [ 14 , 20 ]. This algorithm was found to be suitable for comparison with LC model using both clinical and simulated data [ 14 ]. These approaches quantified the cerebral metabolites N-acetyl aspartate (NAA), Myo-inositol (Mi) and Choline (Cho) and reported as ratios to Cr.

Statistical analysis

All data is expressed as mean±standard deviation (SD), unless state otherwise. All statistical analyzes were carried out on version 17.0 of SPSS (SPSS Inc., Chicago). Comparisons of metabolite ratio in normal tissue and glioma tumor were performed between each software package using a paired sample t-test. The relationship between signal intensity from each of the metabolites obtained by jMRUI and TARQUIN was assessed using a Pearson correlation. P<0.05 was considered as statistically significant.

Results

Comparing the quantitative models jMRUI and TARQUIN

The representative ¹H-MRS spectra from normal tissue and glioma tumor in both programs are shown in Figures 1 and 2. The water suppressed spectra show the peaks at the following chemical shifts: choline (3.2 ppm), creatine (3.01 ppm), NAA (2.0) and mI (3.54). Cho/Cr, NAA/Cr and mI/Cr signal intensities were calculated. The comparison of metabolites between normal group and glioma group has shown in Figure 3. Figure 3 reveals increasing choline in both algorithms as having a similar pattern, and this tendency was found to be statistically significant across the software packages and it is significantly higher in JMRUI. However, both methods could be able to recognize the differences between normal tissue and glioma tumor. Cerebral metabolite ratios (CMRs) for NAA/Cr and Cho/Cr were significantly higher when quantified using the jMRUI software Table 1 displays variation coefficients (CoVs) and confidence intervals of 95 percent for each CMR by algorithm. In general, these were lower for TARQUIN quantified NAA/Cr, Cho/Cr and mI/Cr. When quantified using jMRUI, mI/Cr displayed considerably more variation than the other two cerebral metabolite ratios (CMRs). Generally, both methods were able to detect Glioma metabolism changes compared to healthy tissues, which were specified by the significant ratios of NAA/Cr and Cho/Cr. However, the significant differences of metabolites in the comparison of both methods indicate a change in the results by performing a unique procedure, although the clinical goals for diagnosis of glioma will be achieved in both methods. Table 2 displays the Pearson correlation coefficients for CMR values.

Figure 1. Examples of Magnetic Resonance Spectroscopy (MRS) spectra in java-based graphical for MR user interface packages (jMRUI) model. (A) Normal tissue (B) Glioma tumor.

Figure 2. Examples of Magnetic Resonance Spectroscopy (MRS) spectra in totally automatic robust quantitation in NMR (TARQUIN) model (red) with providing individual peaks (green). (A) Normal tissue (B) Glioma tumor

Figure 3. Changes of signal intensity of Choline/Creatine in java-based graphical for MR user interface packages (jMRUI) and totally automatic robust quantitation in NMR (TARQUIN) software packages in both group.

Discussion

In this research, Glioma tumor data ¹H-MRS were analyzed using two metabolite quantification models namely jMRUI and TARQUIN. This is the first research to evaluate these two models in Glioma tumors. The difference of the quantified values of metabolites was observed by these two models in glioma tumors, which is generally shown as the difference in Cho/Cr ratio. Both models were able to identify the changes caused by Glioma tumor to normal tissue ratio, indicating that both models were able to identify tumors from normal tissue. The metabolites values in the TARQUIN model are estimated to be lower than jMRUI in a study that examined the differences between these two methods in the quantification of metabolites in HIV-infected patients [ 21 ]. In the present research, NAA/Cr and Cho/Cr ratios showed significant differences in quantification by these two models in glioma tumors. The observed differences can be attributed to the fully automated algorithm of the TARQUIN method because in this method the user’s skill in quantification is eliminated. In the JMRUI technique, some steps are done manually for example Water peak suppression and noise removal of the MRS spectrum should be done manually and this user intervention can affect the final quantification of the spectrum, which will influence the test results. The results indicated that the metabolite ratios in Glioma tumors would change using two different models. However, there was no significant change that affects clinical diagnosis and in other words, impairs the diagnosis of glioma tumors from healthy tissue. Further studies will ensure the accuracy and reproducibility of these results.

Conclusion

This study reveals the differentiation of metabolites results from Java-based graphical for MR user interface package (jMRUI) and totally automatic robust quantitation in NMR (TARQUIN) software packages. It can be asserted that both programs can be used for glioma tumor quantification study, but it should be considered that JMRUI software will show higher values in glioma tumor.

Authors’ Contribution

E. Saatchian analyzed the data and wrote the article results, method and discussion. S. Ehsani gathered the magnetic resonance spectroscopy and MRI data and wrote the article introduction. A. Montazerabadi supervised the physique-based analysis of gathered data. All the authors read, modified, and approved the final version of the manuscript.

Ethical Approval

The Ethics Committee of the Faculty of Associated Medical Science at Mashhad University of Medical Science approved this report (Ethic cod: IR.MUMS.MEDICAL.REC.1398.599).

Funding

This study was supported by Mashhad University of Medical Sciences (Grant No: #981148).

Conflict of Interest

None

References

1. Ostrom QT, Gittleman H, Liao P, et al. CBTRUS statistical report: primary brain and central nervous system tumors diagnosed in the United States in 2007–2011. Neuro Oncol. 2014; 16(suppl_4):iv1-63.
2. Horbinski C, Ligon KL, Brastianos P, et al. The medical necessity of advanced molecular testing in the diagnosis and treatment of brain tumor patients. Neuro Oncol. 2019; 21(12):1498-508.
3. Smirniotopoulos JG, Jäger HR. Diseases of the Brain, Head and Neck, Spine 2020–2023: Diagnostic Imaging. Springer: Cham, Switzerland; 2020.
4. Law M, Yang S, Wang H, et al. Glioma grading: sensitivity, specificity, and predictive values of perfusion MR imaging and proton MR spectroscopic imaging compared with conventional MR imaging. AJNR Am J Neuroradiol. 2003; 24(10):1989-98. PubMed
5. Shukla G, Alexander GS, Bakas S, et al. Advanced magnetic resonance imaging in glioblastoma: a review. Chin Clin Oncol. 2017; 6(4)
6. Al-Okaili RN, Krejza J, Wang S, et al. Advanced MR imaging techniques in the diagnosis of intraaxial brain tumors in adults. Radiographics. 2006; 26(Suppl 1):S173-89.
7. Villanueva-Meyer JE, Mabray MC, Cha S. Current Clinical Brain Tumor Imaging. Neurosurgery. 2017; 81(3):397-415.
8. Aydin H, Sipahioğlu S, Oktay NA, et al. The value of proton MR-spectroscopy in the differentiation of brain tumours from non-neoplastic brain lesions. JBR-BTR. 2011; 94(1):1-10. PubMed
9. Horská A, Barker PB. Imaging of brain tumors: MR spectroscopy and metabolic imaging. Neuroimaging Clin N Am. 2010; 20(3):293-310.
10. Soares DP, Law M. Magnetic resonance spectroscopy of the brain: review of metabolites and clinical applications. Clin Radiol. 2009; 64(1):12-21.
11. Dong-Hyun K. Magnetic Resonance Spectroscopy. InTech: Rijeka; 2012.
12. Jabłoński M, Starčuková J, Starčuk Jr Z. Processing tracking in jMRUI software for magnetic resonance spectra quantitation reproducibility assurance. BMC Bioinformatics. 2017; 18(1):56 doi 101186/s12859-017-1459-5. Publisher Full Text | PubMed
13. Wilson M, Reynolds G, Kauppinen RA, et al. A constrained least-squares approach to the automated quantitation of in vivo ¹H magnetic resonance spectroscopy data. Magn Reson Med. 2011; 65(1):1-12.
14. Vanhamme L, Fierro RD, Van Huffel S, De Beer R. Fast Removal of Residual Water in Proton Spectra. J Magn Reson. 1998; 132(2):197-203.
15. Mullins PG, Rowland L, Bustillo J, et al. Reproducibility of 1H-MRS measurements in schizophrenic patients. Magn Reson Med. 2003; 50(4):704-7.
16. Pels P, Ozturk-Isik E, Swanson MG, et al. Quantification of prostate MRSI data by model-based time domain fitting and frequency domain analysis. NMR Biomed. 2006; 19(2):188-97.
17. Vanhamme L, Van Den Boogaart A, Van Huffel S. Improved method for accurate and efficient quantification of MRS data with use of prior knowledge. J Magn Reson. 1997; 129(1):35-43.
18. Van Den Boogaart A. A users guide to the magnetic resonance user interface software package MRUI manual V96.3. Delft, Netherlands; 1997.
19. Poullet JB, Sima DM, Simonetti AW, et al. An automated quantitation of short echo time MRS spectra in an open source software environment: AQSES. NMR Biomed. 2007; 20(5):493-504.
20. Scott J, Underwood J, Garvey LJ, et al. A comparison of two post-processing analysis methods to quantify cerebral metabolites measured via proton magnetic resonance spectroscopy in HIV disease. Br J Radiol. 2016; 89(1060)
21. Lentz MR, Kim WK, Kim H, et al. Alterations in brain metabolism during the first year of HIV infection. J Neurovirol. 2011; 17:220-9.