jSIPRO - analysis tool for magnetic resonance spectroscopic imaging

Comparison of convolutional-neural-networks-based method and LCModel on the quantification of in vivo magnetic resonance spectroscopy

BACKGROUND

Quantification of metabolites concentrations in institutional unit (IU) is important for inter-subject and long-term comparisons in the applications of magnetic resonance spectroscopy (MRS). Recently, deep learning (DL) algorithms have found a variety of applications on the process of MRS data. A quantification strategy compatible to DL base MRS spectral processing method is, therefore, useful.

MATERIALS AND METHODS

This study aims to investigate whether metabolite concentrations quantified using a convolutional neural network (CNN) based method, coupled with a scaling procedure that normalizes spectral signals for CNN input and linear regression, can effectively reflect variations in metabolite concentrations in IU across different brain regions with varying signal-to-noise ratios (SNR) and linewidths (LW). An error index based on standard error (SE) is proposed to indicate the confidence levels associated with metabolite predictions. In vivo MRS spectra were acquired from three brain regions of 43 subjects using a 3T system.

RESULTS

The metabolite concentrations in IU of five major metabolites, quantified using CNN and LCModel, exhibit similar ranges with Pearson's correlation coefficients ranging from 0.24 to 0.78. The SE of the metabolites shows a positive correlation with Cramer-Rao lower bound (CRLB) (r=0.46) and  absolute CRLB (r=0.81), calculated by multiplying CRLBs with the quantified metabolite content.

CONCLUSION

In conclusion, the CNN based method with the proposed scaling procedures can be employed to quantify in vivo MRS spectra and derive metabolites concentrations in IU. The SE can be used as error index, indicating predicted uncertainties for metabolites and sharing information similar to the absolute CRLB.

A Preliminary Study on the Diagnostic Efficacy of Proton Magnetic Resonance Spectroscopy at 3.0T in Rabbit With VX2 Liver Tumor

Background:

To investigate the diagnostic efficacy of choline (Cho) value of magnetic resonance spectroscopy (MRS) in rabbit with VX2 liver tumor via comparative and quantitative analysis with the choline compounds concentration measured by enzyme linked immunosorbent assay (ELISA).

Methods:

MRS was performed on normal liver and VX2 tumor. The Cho value of VX2 tumor was compared with that of normal liver. Tissues were harvested for ELISA to detect the concentrations of acetylcholine (ACh), glycophorophosphygholine (GPC) and phosphochorine (PC). The diagnostic performance of Cho value and concentrations of choline compounds were assessed by receiver operating characteristic (ROC) curve and area under ROC curve (AUC). The specificity and sensitivity were discussed by the maximum Youden’s index.

Results:

The concentration of ACh was obviously higher than that of GPC and PC both in VX2 tumor and normal liver (P < 0.01). Furthermore, the concentration differences among ACh, GPC and PG were the third power of 10. Both the ACh concentration and Cho value of MRS in VX2 tumor were significantly higher than those in normal liver (P < 0.01). The AUC of ACh in VX2 tumor was 0.883, when the cutoff value was 7259000, the sensitivity and specificity of the diagnosis of liver cancer were 94.4% and 77.8%, respectively. The AUC of Cho in VX2 tumor was 0.807, when the cutoff value was 28.35, the sensitivity and specificity of the diagnosis of liver cancer were 83.3% and 77.8%, respectively.

Conclusion:

The change of Cho value in MRS between liver cancer and normal liver was consistent with the changes of concentrations of choline compounds measured by ELISA, especially the change of ACh concentration. The diagnostic efficiency of Cho value and that of choline compounds concentration in liver cancer were extremely similar, with the AUC more than 0.8. We conclude that MRS may be applied as an important, non-invasive biomarker for the diagnosis of liver cancer.

Increased glutamate and deep brain atrophy can predict the severity of multiple sclerosis

OBJECTIVE

In multiple sclerosis (MS), deep grey matter (DGM) atrophy has been recognised as a crucial component of the disease that presents early and it has been associated with disability. Although the precise mechanism underlying grey matter atrophy is unknown, several hypotheses have been postulated. Our previous research pointed to correlations of hypothalamic metabolic alterations with clinical outcomes of MS, therefore we decided to further test the relationship of these alterations with DGM atrophy.

METHODS

We used ¹H-Magnetic Resonance spectroscopy (¹H-MRS) of the hypothalamus to test its metabolites in 26 patients with RRMS and 22 healthy age-matched controls. DGM atrophy was evaluated by simple planimetry of third ventricular width on the hypothalamic level (3VW) in T1 weighted MRI pictures. Metabolite ratios of N-acetyl aspartate (NAA), choline (Cho), glutamate and glutamine (Glx), myo-inositol (mIns) and creatine (Cr) were correlated with Multiple Sclerosis Severity Scale (MSSS) and 3VW.

RESULTS

Metabolite concentrations were compared between patients and controls using multiple regression models allowing for age, 3VW and metabolites. It revealed that the only relevant predictors of MSSS were 3VW and Glx/NAA. At a significance level of P<0.05, a unit increase of 3VW was associated with a 0.35 increase of MSSS, for a typical value of Glx/NAA; P value 0.0039. A unit increase of Glx/NAA was associated with a 0.93 increase of MSSS, for a typical value of atrophy; P value 0.090. There were significant linear correlations between Glx/Cr and MSSS, Glx/NAA and MSSS, and between mIns/NAA and 3VW.

CONCLUSIONS

The results suggest that both NAA and Glx are associated with neurodegeneration of hypothalamic DGM and severe disease course. Glx related ¹H-MRS parameters seem to be superior to other metabolites in determining disease burden, independently of otherwise powerful 3VW planimetry. Significantly increased mIns/NAA in MS patients compared to controls point to gliosis, which parallels the atrophy of hypothalamic DGM.

Post-WBRT cognitive impairment and hippocampal neuronal depletion measured by in vivo metabolic MR spectroscopy: Results of prospective investigational study

BACKGROUND AND PURPOSE

The aim of this prospective study is to evaluate post-whole brain radiotherapy (WBRT) changes in hippocampal concentration of N-acetylaspartate (h-tNAA) as a marker of neuronal loss and to correlate those changes to neurocognitive function.

MATERIAL AND METHODS

Thirty-five patients with brain metastases underwent baseline single slice multi-voxel MR spectroscopy (MRS) examination for measurement of hippocampal h-tNAA together with baseline battery of neurocognitive tests focused on memory (Auditory Verbal Learning Test and Brief Visuospatial Memory Test - Revised) as well as quality of life questionnaires (EORTC QLQ-C30 a EORTC QLQ-BN20). Eighteen patients completed follow-up evaluation four months after standard WBRT (2 laterolateral fields, 10×3.0Gy, 6MV photons) and were included in this analysis. MRS and cognitive examinations were repeated and compared to baseline measurements.

RESULTS

Statistically significant decreases in h-tNAA were observed in the right (8.52-7.42mM; -12.9%, 95%CI: -7.6 to -16.4%) as well as in the left hippocampus (8.64-7.60mM; -12%, 95%CI: -7.9 to -16.2%). Statistically significant decline was observed in all AVLT and BVMT-R subtests with exception of AVLT_Recognition. Quality of life declined after WBRT (mean Δ -14.1±20.3 points in transformed 0-100 point scale; p=0.018) with no correlation to changes in hippocampal metabolite concentrations. Moderate positive correlation was observed between left h-tNAA concentration decrease and AVLT_TR decline (r=+0.32; p=0.24) as well as with AVLT_DR (r=+0.33; p=0.22) decline. Changes in right h-tNAA/Cr negatively correlated with AVLT_DR (r=-0.48; p=0.061). No correlation between right hippocampus h-tNAA and memory decline (AVLT) was observed.

CONCLUSIONS

Our results suggest hippocampal NAA concentrations decline after WBRT and MRS may be a useful biomarker for monitoring neuronal loss after radiotherapy.

The aging effect on prostate metabolite concentrations measured by 1H MR spectroscopy

OBJECTIVE

The effects of aging, magnetic field and the voxel localization on measured concentrations of citrate (Cit), creatine (Cr), cholines (Cho) and polyamines (PA) in a healthy prostate were evaluated.

MATERIALS AND METHODS

36 examinations at both 1.5T and 3T imagers of 52 healthy subjects aged 19-71 years were performed with PRESS 3D-CSI sequences (TE = 120 and 145 ms). Concentrations in laboratory units and their ratios to citrate were calculated using the LCModel technique. Absolute concentrations were also obtained after the application of correction coefficients. Statistical analysis was performed using a robust linear mixed effects model.

RESULTS

Significant effects of aging, the magnetic field strength and the voxel position in central (CZ) or peripheral (PZ) zones on all measured metabolites were found. The concentrations (mmol/kg wet tissue) including prediction intervals in a range of 20-70 years were found: Cit: 7.9-17.2; Cho: 1.4-1.7; Cr: 2.8-2.5; PA (as spermine): 0.6-2.1 at 3T in CZ. In PZ, the concentrations were higher by about 10 % as compared to CZ.

CONCLUSION

Increasing citrate and spermine concentrations with age are significant and correlate well with a recently described increase of zinc in the prostate. These findings should be considered in decision-making if the values obtained from a subject are in the range of control values.

Advanced MRI increases the diagnostic accuracy of recurrent glioblastoma: Single institution thresholds and validation of MR spectroscopy and diffusion weighted MR imaging

The accurate identification of glioblastoma progression remains an unmet clinical need. The aim of this prospective single-institutional study is to determine and validate thresholds for the main metabolite concentrations obtained by MR spectroscopy (MRS) and the values of the apparent diffusion coefficient (ADC) to enable distinguishing tumor recurrence from pseudoprogression. Thirty-nine patients after the standard treatment of a glioblastoma underwent advanced imaging by MRS and ADC at the time of suspected recurrence — median time to progression was 6.7 months. The highest significant sensitivity and specificity to call the glioblastoma recurrence was observed for the total choline (tCho) to total N-acetylaspartate (tNAA) concentration ratio with the threshold ≥ 1.3 (sensitivity 100.0% and specificity 94.7%). The ADCmean value higher than 1313 × 10^− 6 mm²/s was associated with the pseudoprogression (sensitivity 98.3%, specificity 100.0%). The combination of MRS focused on the tCho/tNAA concentration ratio and the ADCmean value represents imaging methods applicable to early non-invasive differentiation between a glioblastoma recurrence and a pseudoprogression. However, the institutional definition and validation of thresholds for differential diagnostics is needed for the elimination of setup errors before implementation of these multimodal imaging techniques into clinical practice, as well as into clinical trials.

•

For an effective salvage treatment, an accurate diagnosis of GBM recurrence is essential.

•

The standard structural MRI has limited sensitivity and specificity to distinguish GBM progression.

•

GBM recurrence is characterized by the ADCmean value ≤ 1313 × 10^− 6 mm²/s and the tCho/tNAA ratio ≥ 1.3.

•

An institutional definition of thresholds is needed, if advanced imaging should be used accurately in clinical practice.

A novel fMRI group data analysis method based on data-driven reference extracting from group subjects

Group-independent component analysis (GICA) is a well-established blind source separation technique that has been widely applied to study multi-subject functional magnetic resonance imaging (fMRI) data. The group-independent components (GICs) represent the commonness of all of the subjects in the group. Similar to independent component analysis on the single-subject level, the performance of GICA can be improved for multi-subject fMRI data analysis by incorporating a priori information; however, a priori information is not always considered while looking for GICs in existing GICA methods, especially when no obvious or specific knowledge about an unknown group is available. In this paper, we present a novel method to extract the group intrinsic reference from all of the subjects of the group and then incorporate it into the GICA extraction procedure. Comparison experiments between FastICA and GICA with intrinsic reference (GICA-IR) are implemented on the group level with regard to the simulated, hybrid and real fMRI data. The experimental results show that the GICs computed by GICA-IR have a higher correlation with the corresponding independent component of each subject in the group, and the accuracy of activation regions detected by GICA-IR was also improved. These results have demonstrated the advantages of the GICA-IR method, which can better reflect the commonness of the subjects in the group.

Minipig model of Huntington's disease: ¹H magnetic resonance spectroscopy of the brain

Huntington's disease (HD) is an inherited autosomal neurodegenerative disorder affecting predominantly the brain, characterized by motor dysfunctions, behavioral and cognitive disturbances. The aim of this study was to determine changes in the brain of transgenic minipigs before HD onset using (1)H magnetic resonance (MR) spectroscopy. Measurements were performed on a 3 T MR scanner using a single voxel spectroscopy sequence for spectra acquisition in the white matter and chemical shift imaging sequence for measurement in the striatum, hippocampus and thalamus. A decrease of (phospho)creatine (tCr) concentration was found only in the thalamus (p=0.002) of transgenic minipigs, nevertheless we found significant changes in metabolite ratios. Increase of the ratio choline compounds (tCho)/tCr was found in all examined areas: striatum (p=0.010), thalamus (p=0.011) as well as hippocampus (p=0.027). The ratio N-acetylaspartate+N-acetylaspartylglutamate (tNAA)/tCr (p=0.043) and glutamate+glutamine (Glx)/tCr (p=0.039) was elevated in the thalamus, the ratio myo-inositol (Ins)/tCr (p=0.048) was significantly increased in the hippocampus. No significant differences were observed in the metabolite concentrations in the white matter, however we found significant increase of ratios tNAA/tCr (p=0.018) and tCho/tCr (p=0.003) ratios in transgenic boars. We suppose that the majority of the observed changes are predominantly related to changes in energy metabolism caused by decrease of tCr.

Hippocampal proton MR spectroscopy as a novel approach in the assessment of radiation injury and the correlation to neurocognitive function impairment: initial experiences

BACKGROUND

The hippocampus is considered as the main radiosensitive brain structure responsible for postradiotherapy cognitive decline. We prospectively assessed correlation of memory change to hippocampal N-acetylaspartate (h-tNAA) concentration, a neuronal density and viability marker, by (1)H-MR spectroscopy focused on the hippocampus.

METHODS

Patients with brain metastases underwent whole brain radiotherapy (WBRT) to a dose of 30 Gy in ten fractions daily. Pre-radiotherapy (1)H-MR spectroscopy focused on the h-tNAA concentration and memory testing was performed. Memory was evaluated by Auditory Verbal Learning Test (AVLT) and Brief Visuospatial Memory Test-Revised (BVMT-R). Total recall, recognition and delayed recall were reported. The both investigation procedures were repeated 4 months after WBRT and the h-tNAA and memory changes were correlated.

RESULTS

Of the 20 patients, ten passed whole protocol. The h-tNAA concentration significantly decreased from pre-WBRT 8.9, 8.86 and 8.88 [mM] in the right, left and both hippocampi to 7.16, 7.65 and 7.4 after WBRT, respectively. In the memory tests a significant decrease was observed in AVLT total-recall, BVMT-R total-recall and BVMT-R delayed-recall. Weak to moderate correlations were observed between left h-tNAA and AVLT recognition and all BVMT-R subtests and between the right h-tNAA and AVLT total-recall.

CONCLUSIONS

A significant decrease in h-tNAA after WBRT was proven by (1)H-MR spectroscopy as a feasible method for the in vivo investigation of radiation injury. Continuing patient recruitment focusing on other cognitive tests and metabolites is needed.

The Diagnostic Ability of Follow-Up Imaging Biomarkers after Treatment of Glioblastoma in the Temozolomide Era: Implications from Proton MR Spectroscopy and Apparent Diffusion Coefficient Mapping

Objective. To prospectively determine institutional cut-off values of apparent diffusion coefficients (ADCs) and concentration of tissue metabolites measured by MR spectroscopy (MRS) for early differentiation between glioblastoma (GBM) relapse and treatment-related changes after standard treatment. Materials and Methods. Twenty-four GBM patients who received gross total resection and standard adjuvant therapy underwent MRI examination focusing on the enhancing region suspected of tumor recurrence. ADC maps, concentrations of N-acetylaspartate, choline, creatine, lipids, and lactate, and metabolite ratios were determined. Final diagnosis as determined by biopsy or follow-up imaging was correlated to the results of advanced MRI findings. Results. Eighteen (75%) and 6 (25%) patients developed tumor recurrence and pseudoprogression, respectively. Mean time to radiographic progression from the end of chemoradiotherapy was 5.8 ± 5.6 months. Significant differences in ADC and MRS data were observed between those with progression and pseudoprogression. Recurrence was characterized by N-acetylaspartate ≤ 1.5 mM, choline/N-acetylaspartate ≥ 1.4 (sensitivity 100%, specificity 91.7%), N-acetylaspartate/creatine ≤ 0.7, and ADC ≤ 1300 × 10⁻⁶ mm²/s (sensitivity 100%, specificity 100%). Conclusion. Institutional validation of cut-off values obtained from advanced MRI methods is warranted not only for diagnosis of GBM recurrence, but also as enrollment criteria in salvage clinical trials and for reporting of outcomes of initial treatment.

The Relationships Between Quantitative MR Parameters in Hippocampus in Healthy Subjects and Patients With Temporal Lobe Epilepsy

We introduce a new magnetic resonance (MR) method based on a pixel-by-pixel image processing to examine relationships between metabolic and structural processes in the pathologic hippocampus. The method was tested for lateralization of the epileptogenic zone in patients with temporal lobe epilepsy (TLE). Twenty patients with drug-resistant TLE and fifteen healthy controls were examined at 3T. The measurement protocol contained T2-weighted MR images, spectroscopic imaging, diffusion tensor imaging and T2 relaxometry. Correlations between quantitative MR parameters were calculated on a pixel-by-pixel basis using the CORIMA program which enables automated pixel identification in the normal tissue according to control data. All MR parameters changed in the anteroposterior direction in the hippocampus and correlation patterns and their slopes differed between patients and controls. Combinations of T2 relaxation times with metabolite values represent the best biomarkers of the epileptogenic zone. Correlations with mean diffusivity did not provide sufficiently accurate results due to diffusion image distortions. Quantitative MR analysis non-invasively provides a detailed description of hippocampal pathology and may represent complementary tool to the standard clinical protocol. However, the automated processing should be carefully monitored in order to avoid possible errors caused by MR artifacts.