Strengths and weaknesses of 1.5T and 3T MRS data in brain glioma classification

Using Single-Voxel Magnetic Resonance Spectroscopy Data Acquired at 1.5T to Classify Multivoxel Data at 3T: A Proof-of-Concept Study

In vivo magnetic resonance spectroscopy (MRS) has two modalities, single-voxel (SV) and multivoxel (MV), in which one or more contiguous grids of SVs are acquired.

PURPOSE

To test whether MV grids can be classified with models trained with SV.

METHODS

Retrospective study. Training dataset: Multicenter multiformat SV INTERPRET, 1.5T. Testing dataset: MV eTumour, 3T. Two classification tasks were completed: 3-class (meningioma vs. aggressive vs. normal) and 4-class (meningioma vs. low-grade glioma vs. aggressive vs. normal). Five different methods were tested for feature selection. The classification was implemented using linear discriminant analysis (LDA), random forest, and support vector machines. The evaluation was completed with balanced error rate (BER) and area under the curve (AUC) on both sets. The accuracy in class prediction was calculated by developing a solid tumor index (STI) and segmentation accuracy with the Dice score.

RESULTS

The best method was sequential forward feature selection combined with LDA, with AUCs = 0.95 (meningioma), 0.89 (aggressive), 0.82 (low-grade glioma), and 0.82 (normal). STI was 66% (4-class task) and 71% (3-class task) because two cases failed completely and two more had suboptimal STI as defined by us.

DISCUSSION

The reasons for failure in the classification of the MV test set were related to the presence of artifacts.

Automatic Brain Tumor Classification via Lion Plus Dragonfly Algorithm

Denoising, skull stripping, segmentation, feature extraction, and classification are five important processes in this paper's development of a brain tumor classification model. The brain tumor image will be imposed first using the entropy-based trilateral filter to de-noising and this image is imposed to skull stripping by means of morphological partition and Otsu thresholding. Adaptive contrast limited fuzzy adaptive histogram equalization (CLFAHE) is also used in the segmentation process. The gray-level co-occurrence matrix (GLCM) characteristics are derived from the segmented image. The collected GLCM features are used in a hybrid classifier that combines the neural network (NN) and deep belief network (DBN) ideas. As an innovation, the hidden neurons of the two classifiers are modified ideally to improve the prediction model's accuracy. The hidden neurons are optimized using a unique hybrid optimization technique known as lion with dragonfly separation update (L-DSU), which integrates the approaches from both DA and LA. Finally, the suggested model's performance is compared to that of the standard models concerning certain performance measures.

Intraoperative 3 T MRI is more correlative to residual disease extent than early postoperative MRI

PURPOSE

Extent of resection of low grade glioma (LGG) is an important prognostic variable, and may influence decisions regarding adjuvant therapy in certain patient populations. Immediate postoperative magnetic resonance image (MRI) is the mainstay for assessing residual tumor. However, previous studies have suggested that early postoperative MRI fluid-attenuated inversion recovery (FLAIR) (within 48 h) may overestimate residual tumor volume in LGG. Intraoperative magnetic resonance imaging (iMRI) without subsequent resection may more accurately assess residual tumor. Consistency in MRI techniques and utilization of higher magnet strengths may further improve both comparisons between MRI studies performed at different time points as well as the specificity of MRI findings to identify residual tumor. To evaluate the utility of 3 T iMRI in the imaging of LGG, we volumetrically analyzed intraoperative, early, and late (~ 3 months after surgery) postoperative MRIs after resection of LGG.

METHODS

A total of 32 patients with LGG were assessed retrospectively. Residual tumor was defined as hyperintense T2 signal on FLAIR. Volumetric assessment was performed with intraoperative, early, and late postoperative FLAIR via TeraRecon iNtuition.

RESULTS

Perilesional FLAIR parenchymal abnormality volumes were significantly different comparing intraoperative and early postoperative MRI (2.17 ± 0.45 cm³ vs. 5.47 ± 1.07 cm³, respectively (p = 0.0002)). A significant difference of perilesional FLAIR parenchymal abnormality volumes was also found comparing early and late postoperative MRI (5.47 ± 1.07 cm³ vs. 3.22 ± 0.64 cm³, respectively (p = 0.0001)). There was no significant difference between intraoperative and late postoperative Perilesional FLAIR parenchymal abnormality volumes.

CONCLUSIONS

Intraoperative 3 T MRI without further resection appears to better reflect the volume of residual tumor in LGG compared with early postoperative 3 T MRI. Early postoperative MRI may overestimate residual tumor. As such, intraoperative MRI performed after completion of tumor resection may be more useful for making decisions regarding adjuvant therapy.

Noninvasively Evaluating the Grading of Glioma by Multiparametric Magnetic Resonance Imaging

RATIONALE AND OBJECTIVE

To investigate the performance of multi-parametric magnetic resonance imaging (MRI) for glioma grading.

MATERIALS AND METHODS

Seventy consecutive patients with histopathologically confirmed glioma were retrospectively evaluated by conventional MRI, dynamic susceptibility-weighted contrast-enhanced, multiple diffusion-weighted imaging signal models including mono-exponential, bi-exponential, stretched exponential, and diffusion kurtosis imaging. One-way analysis of variance and independent-samples t test were used to compare the MR parameter values between low and high grades as well as among all grades of glioma. Receiver operating characteristic analysis, Spearman's correlation analysis, and binary logistic regression analysis were used to assess their diagnostic performance.

RESULTS

The diagnostic performance (the optimal thresholds, area under the receiver operating characteristic curve, sensitivity, and specificity) was achieved with normalized relative cerebral blood flow (rCBV) (2.240 ml/100 g, 0.844, 87.8%, and 75.9%, respectively), mean kurtosis (MK) (0.471, 0.873, 92.7%, and 79.3%), and water molecular diffusion heterogeneity index (α) (1.064, 0.847, 79.3% and 78.0%) for glioma grading. There were positive correlations between rCBV and MK and the tumor grades and negative correlations between α and the tumor grades (p < 0.01). The parameter of α yielded a diagnostic accuracy of 85.3%, the combination of MK and α yielded a diagnostic accuracy of 89.7%, while the combination of rCBV, MK, and α were more accurate (94.2%) in predicting tumor grade.

CONCLUSION

The most accurate parameters were rCBV, MK, and α in dynamic susceptibility-weighted contrast, diffusion kurtosis imaging, and Multi-b diffusion-weighted imaging for glioma grading, respectively. Multiparametric MRI can increase the accuracy of glioma grading.

Integrating diffusion kurtosis imaging, dynamic susceptibility-weighted contrast-enhanced MRI, and short echo time chemical shift imaging for grading gliomas

BACKGROUND

We assessed the diagnostic accuracy of diffusion kurtosis imaging (DKI), dynamic susceptibility-weighted contrast-enhanced (DSC) MRI, and short echo time chemical shift imaging (CSI) for grading gliomas.

METHODS

In this prospective study, 35 patients with cerebral gliomas underwent DKI, DSC, and CSI on a 3 T MR scanner. Diffusion parameters were mean diffusivity (MD), fractional anisotropy, and mean kurtosis (MK). Perfusion parameters were mean relative regional cerebral blood volume (rrCBV), mean relative regional cerebral blood flow (rrCBF), mean transit time, and relative decrease ratio (rDR). The diffusion and perfusion parameters along with 12 CSI metabolite ratios were compared among 22 high-grade gliomas and 14 low-grade gliomas (Mann-Whitney U-test, P < .05). Classification accuracy was determined with a linear discriminant analysis for each MR modality independently. Furthermore, the performance of a multimodal analysis is reported, using a decision-tree rule combining the statistically significant DKI, DSC-MRI, and CSI parameters with the lowest P-value. The proposed classifiers were validated on a set of subsequently acquired data from 19 clinical patients.

RESULTS

Statistically significant differences among tumor grades were shown for MK, MD, mean rrCBV, mean rrCBF, rDR, lipids over total choline, lipids over creatine, sum of myo-inositol, and sum of creatine. DSC-MRI proved to be the modality with the best performance when comparing modalities individually, while the multimodal decision tree proved to be most accurate in predicting tumor grade, with a performance of 86%.

CONCLUSIONS

Combining information from DKI, DSC-MRI, and CSI increases diagnostic accuracy to differentiate low- from high-grade gliomas, possibly providing diagnosis for the individual patient.

Tumour Relapse Prediction Using Multiparametric MR Data Recorded during Follow-Up of GBM Patients

Purpose. We have focused on finding a classifier that best discriminates between tumour progression and regression based on multiparametric MR data retrieved from follow-up GBM patients. Materials and Methods. Multiparametric MR data consisting of conventional and advanced MRI (perfusion, diffusion, and spectroscopy) were acquired from 29 GBM patients treated with adjuvant therapy after surgery over a period of several months. A 27-feature vector was built for each time point, although not all features could be obtained at all time points due to missing data or quality issues. We tested classifiers using LOPO method on complete and imputed data. We measure the performance by computing BER for each time point and wBER for all time points. Results. If we train random forests, LogitBoost, or RobustBoost on data with complete features, we can differentiate between tumour progression and regression with 100% accuracy, one time point (i.e., about 1 month) earlier than the date when doctors had put a label (progressive or responsive) according to established radiological criteria. We obtain the same result when training the same classifiers solely on complete perfusion data. Conclusions. Our findings suggest that ensemble classifiers (i.e., random forests and boost classifiers) show promising results in predicting tumour progression earlier than established radiological criteria and should be further investigated.

A novel semi-supervised methodology for extracting tumor type-specific MRS sources in human brain data

Background

The clinical investigation of human brain tumors often starts with a non-invasive imaging study, providing information about the tumor extent and location, but little insight into the biochemistry of the analyzed tissue. Magnetic Resonance Spectroscopy can complement imaging by supplying a metabolic fingerprint of the tissue. This study analyzes single-voxel magnetic resonance spectra, which represent signal information in the frequency domain. Given that a single voxel may contain a heterogeneous mix of tissues, signal source identification is a relevant challenge for the problem of tumor type classification from the spectroscopic signal.

Methodology/Principal Findings

Non-negative matrix factorization techniques have recently shown their potential for the identification of meaningful sources from brain tissue spectroscopy data. In this study, we use a convex variant of these methods that is capable of handling negatively-valued data and generating sources that can be interpreted as tumor class prototypes. A novel approach to convex non-negative matrix factorization is proposed, in which prior knowledge about class information is utilized in model optimization. Class-specific information is integrated into this semi-supervised process by setting the metric of a latent variable space where the matrix factorization is carried out. The reported experimental study comprises 196 cases from different tumor types drawn from two international, multi-center databases. The results indicate that the proposed approach outperforms a purely unsupervised process by achieving near perfect correlation of the extracted sources with the mean spectra of the tumor types. It also improves tissue type classification.

Conclusions/Significance

We show that source extraction by unsupervised matrix factorization benefits from the integration of the available class information, so operating in a semi-supervised learning manner, for discriminative source identification and brain tumor labeling from single-voxel spectroscopy data. We are confident that the proposed methodology has wider applicability for biomedical signal processing.