A multi-centre, web-accessible and quality control-checked database of in vivo MR spectra of brain tumour patients

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.

Extraction of artefactual MRS patterns from a large database using non-negative matrix factorization

Despite the success of automated pattern recognition methods in problems of human brain tumor diagnostic classification, limited attention has been paid to the issue of automated data quality assessment in the field of MRS for neuro-oncology. Beyond some early attempts to address this issue, the current standard in practice is MRS quality control through human (expert-based) assessment. One aspect of automatic quality control is the problem of detecting artefacts in MRS data. Artefacts, whose variety has already been reviewed in some detail and some of which may even escape human quality control, have a negative influence in pattern recognition methods attempting to assist tumor characterization. The automatic detection of MRS artefacts should be beneficial for radiology as it guarantees more reliable tumor characterizations, as well as the development of more robust pattern recognition-based tumor classifiers and more trustable MRS data processing and analysis pipelines. Feature extraction methods have previously been used to help distinguishing between good and bad quality spectra to apply subsequent supervised pattern recognition techniques. In this study, we apply feature extraction differently and use a variant of a method for blind source separation, namely Convex Non-Negative Matrix Factorization, to unveil MRS signal sources in a completely unsupervised way. We hypothesize that, while most sources will correspond to the different tumor patterns, some of them will reflect signal artefacts. The experimental work reported in this paper, analyzing a combined short and long echo time ¹ H-MRS database of more than 2000 spectra acquired at 1.5T and corresponding to different tumor types and other anomalous masses, provides a first proof of concept that points to the possible validity of this approach.

Robust Conditional Independence maps of single-voxel Magnetic Resonance Spectra to elucidate associations between brain tumours and metabolites

The aim of the paper is two-fold. First, we show that structure finding with the PC algorithm can be inherently unstable and requires further operational constraints in order to consistently obtain models that are faithful to the data. We propose a methodology to stabilise the structure finding process, minimising both false positive and false negative error rates. This is demonstrated with synthetic data. Second, to apply the proposed structure finding methodology to a data set comprising single-voxel Magnetic Resonance Spectra of normal brain and three classes of brain tumours, to elucidate the associations between brain tumour types and a range of observed metabolites that are known to be relevant for their characterisation. The data set is bootstrapped in order to maximise the robustness of feature selection for nominated target variables. Specifically, Conditional Independence maps (CI-maps) built from the data and their derived Bayesian networks have been used. A Directed Acyclic Graph (DAG) is built from CI-maps, being a major challenge the minimization of errors in the graph structure. This work presents empirical evidence on how to reduce false positive errors via the False Discovery Rate, and how to identify appropriate parameter settings to improve the False Negative Reduction. In addition, several node ordering policies are investigated that transform the graph into a DAG. The obtained results show that ordering nodes by strength of mutual information can recover a representative DAG in a reasonable time, although a more accurate graph can be recovered using a random order of samples at the expense of increasing the computation time.

A metabolic database for biomedical studies of biopsy specimens by high-resolution magic angle spinning nuclear MR: a qualitative and quantitative tool

PURPOSE

The aim of this study is to generate a metabolic database for biomedical studies of biopsy specimens by high-resolution magic angle spinning (HRMAS) nuclear MR (NMR).

METHODS

Seventy-six metabolites, classically found in human biopsy samples, were prepared in aqueous solution at a known concentration and analyzed by HRMAS NMR. The spectra were recorded under the same conditions as the ones used for the analysis of biopsy specimens routinely performed in our hospital.

RESULTS

For each metabolite, a complete set of NMR spectra (1D ¹ H, 1D ¹ H-CPMG, 2D J-Resolved, 2D TOCSY, and 2D ¹ H-¹³ C HSQC) was recorded at 500 MHz and 277 K. All spectra were manually assigned using the information contained in the different spectra and existing databases. Experiments to measure the T₁ and the T₂ of the different protons present in the 76 metabolites were also recorded.

CONCLUSION

This new HRMAS metabolic database is a useful tool for all scientists working on human biopsy specimens, particularly in the field of oncology. It will make the identification of metabolites in biopsy specimens faster and more reliable. Additionally, the knowledge of the T₁ and T₂ values will allow to obtain a more accurate quantification of the metabolites present in biopsy specimens.

Quality of clinical brain tumor MR spectra judged by humans and machine learning tools

PURPOSE

To investigate and compare human judgment and machine learning tools for quality assessment of clinical MR spectra of brain tumors.

METHODS

A very large set of 2574 single voxel spectra with short and long echo time from the eTUMOUR and INTERPRET databases were used for this analysis. Original human quality ratings from these studies as well as new human guidelines were used to train different machine learning algorithms for automatic quality control (AQC) based on various feature extraction methods and classification tools. The performance was compared with variance in human judgment.

RESULTS

AQC built using the RUSBoost classifier that combats imbalanced training data performed best. When furnished with a large range of spectral and derived features where the most crucial ones had been selected by the TreeBagger algorithm it showed better specificity (98%) in judging spectra from an independent test-set than previously published methods. Optimal performance was reached with a virtual three-class ranking system.

CONCLUSION

Our results suggest that feature space should be relatively large for the case of MR tumor spectra and that three-class labels may be beneficial for AQC. The best AQC algorithm showed a performance in rejecting spectra that was comparable to that of a panel of human expert spectroscopists. Magn Reson Med 79:2500-2510, 2018. © 2017 International Society for Magnetic Resonance in Medicine.

Big GABA: Edited MR spectroscopy at 24 research sites

Magnetic resonance spectroscopy (MRS) is the only biomedical imaging method that can noninvasively detect endogenous signals from the neurotransmitter γ-aminobutyric acid (GABA) in the human brain. Its increasing popularity has been aided by improvements in scanner hardware and acquisition methodology, as well as by broader access to pulse sequences that can selectively detect GABA, in particular J-difference spectral editing sequences. Nevertheless, implementations of GABA-edited MRS remain diverse across research sites, making comparisons between studies challenging. This large-scale multi-vendor, multi-site study seeks to better understand the factors that impact measurement outcomes of GABA-edited MRS. An international consortium of 24 research sites was formed. Data from 272 healthy adults were acquired on scanners from the three major MRI vendors and analyzed using the Gannet processing pipeline. MRS data were acquired in the medial parietal lobe with standard GABA+ and macromolecule- (MM-) suppressed GABA editing. The coefficient of variation across the entire cohort was 12% for GABA+ measurements and 28% for MM-suppressed GABA measurements. A multilevel analysis revealed that most of the variance (72%) in the GABA+ data was accounted for by differences between participants within-site, while site-level differences accounted for comparatively more variance (20%) than vendor-level differences (8%). For MM-suppressed GABA data, the variance was distributed equally between site- (50%) and participant-level (50%) differences. The findings show that GABA+ measurements exhibit strong agreement when implemented with a standard protocol. There is, however, increased variability for MM-suppressed GABA measurements that is attributed in part to differences in site-to-site data acquisition. This study's protocol establishes a framework for future methodological standardization of GABA-edited MRS, while the results provide valuable benchmarks for the MRS community.

Classification of brain tumours from MR spectra: the INTERPRET collaboration and its outcomes

The INTERPRET project was a multicentre European collaboration, carried out from 2000 to 2002, which developed a decision-support system (DSS) for helping neuroradiologists with no experience of MRS to utilize spectroscopic data for the diagnosis and grading of human brain tumours. INTERPRET gathered a large collection of MR spectra of brain tumours and pseudo-tumoural lesions from seven centres. Consensus acquisition protocols, a standard processing pipeline and strict methods for quality control of the aquired data were put in place. Particular emphasis was placed on ensuring the diagnostic certainty of each case, for which all cases were evaluated by a clinical data validation committee. One outcome of the project is a database of 304 fully validated spectra from brain tumours, pseudotumoural lesions and normal brains, along with their associated images and clinical data, which remains available to the scientific and medical community. The second is the INTERPRET DSS, which has continued to be developed and clinically evaluated since the project ended. We also review here the results of the post-INTERPRET period. We evaluate the results of the studies with the INTERPRET database by other consortia or research groups. A summary of the clinical evaluations that have been performed on the post-INTERPRET DSS versions is also presented. Several have shown that diagnostic certainty can be improved for certain tumour types when the INTERPRET DSS is used in conjunction with conventional radiological image interpretation. About 30 papers concerned with the INTERPRET single-voxel dataset have so far been published. We discuss stengths and weaknesses of the DSS and the lessons learned. Finally we speculate on how the INTERPRET concept might be carried into the future.

Multicentre evaluation of the INTERPRET decision support system 2.0 for brain tumour classification

In a previous study, we have shown the added value of (1) H MRS for the neuroradiological characterisation of adult human brain tumours. In that study, several methods of MRS analysis were used, and a software program, the International Network for Pattern Recognition of Tumours Using Magnetic Resonance Decision Support System 1.0 (INTERPRET DSS 1.0), with a short-TE classifier, provided the best results. Since then, the DSS evolved into a version 2.0 that contains an additional long-TE classifier. This study has two objectives. First, to determine whether clinicians with no experience of spectroscopy are comparable with spectroscopists in the use of the system, when only minimum training in the use of the system was given. Second, to assess whether or not a version with another TE is better than the initial version. We undertook a second study with the same cases and nine evaluators to assess whether the diagnostic accuracy of DSS 2.0 was comparable with the values obtained with DSS 1.0. In the second study, the analysis protocol was flexible in comparison with the first one to mimic a clinical environment. In the present study, on average, each case required 5.4 min by neuroradiologists and 9 min by spectroscopists for evaluation. Most classes and superclasses of tumours gave the same results as with DSS 1.0, except for astrocytomas of World Health Organization (WHO) grade III, in which performance measured as the area under the curve (AUC) decreased: AUC = 0.87 (0.72-1.02) with DSS 1.0 and AUC = 0.62 (0.55-0.70) with DSS 2.0. When analysing the performance of radiologists and spectroscopists with respect to DSS 1.0, the results were the same for most classes. Having data with two TEs instead of one did not affect the results of the evaluation.

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.

Extracting MRS discriminant functional features of brain tumors

The current challenge in automatic brain tumor classification based on MRS is the improvement of the robustness of the classification models that explicitly account for the probable breach of the independent and identically distributed conditions in the MRS data points. To contribute to this purpose, a new algorithm for the extraction of discriminant MRS features of brain tumors based on a functional approach is presented. Functional data analysis based on region segmentation (RSFDA) is based on the functional data analysis formalism using nonuniformly distributed B splines according to spectral regions that are highly correlated. An exhaustive characterization of the method is presented in this work using controlled and real scenarios. The performance of RSFDA was compared with other widely used feature extraction methods. In all simulated conditions, RSFDA was proven to be stable with respect to the number of variables selected and with respect to the classification performance against noise and baseline artifacts. Furthermore, with real multicenter datasets classification, RSFDA and peak integration (PI) obtained better performance than the other feature extraction methods used for comparison. Other advantages of the method proposed are its usefulness in selecting the optimal number of features for classification and its simplified functional representation of the spectra, which contributes to highlight the discriminative regions of the MR spectrum for each classification task.

Strategies for annotation and curation of translational databases: the eTUMOUR project

The eTUMOUR (eT) multi-centre project gathered in vivo and ex vivo magnetic resonance (MR) data, as well as transcriptomic and clinical information from brain tumour patients, with the purpose of improving the diagnostic and prognostic evaluation of future patients. In order to carry this out, among other work, a database—the eTDB—was developed. In addition to complex permission rules and software and management quality control (QC), it was necessary to develop anonymization, processing and data visualization tools for the data uploaded. It was also necessary to develop sophisticated curation strategies that involved on one hand, dedicated fields for QC-generated meta-data and specialized queries and global permissions for senior curators and on the other, to establish a set of metrics to quantify its contents. The indispensable dataset (ID), completeness and pairedness indices were set. The database contains 1317 cases created as a result of the eT project and 304 from a previous project, INTERPRET. The number of cases fulfilling the ID was 656. Completeness and pairedness were heterogeneous, depending on the data type involved.

Accurate classification of childhood brain tumours by in vivo ¹H MRS - a multi-centre study

AIMS

To evaluate the accuracy of single-voxel Magnetic Resonance Spectroscopy ((1)H MRS) as a non-invasive diagnostic aid for paediatric brain tumours in a multi-national study. Our hypotheses are (1) that automated classification based on (1)H MRS provides an accurate non-invasive diagnosis in multi-centre datasets and (2) using a protocol which increases the metabolite information improves the diagnostic accuracy.

METHODS

Seventy-eight patients under 16 years old with histologically proven brain tumours from 10 international centres were investigated. Discrimination of 29 medulloblastomas, 11 ependymomas and 38 pilocytic astrocytomas (PILOAs) was evaluated. Single-voxel MRS was undertaken prior to diagnosis (1.5 T Point-Resolved Spectroscopy (PRESS), Proton Brain Exam (PROBE) or Stimulated Echo Acquisition Mode (STEAM), echo time (TE) 20-32 ms and 135-136 ms). MRS data were processed using two strategies, determination of metabolite concentrations using TARQUIN software and automatic feature extraction with Peak Integration (PI). Linear Discriminant Analysis (LDA) was applied to this data to produce diagnostic classifiers. An evaluation of the diagnostic accuracy was performed based on resampling to measure the Balanced Accuracy Rate (BAR).

RESULTS

The accuracy of the diagnostic classifiers for discriminating the three tumour types was found to be high (BAR 0.98) when a combination of TE was used. The combination of both TEs significantly improved the classification performance (p<0.01, Tukey's test) compared with the use of one TE alone. Other tumour types were classified accurately as glial or primitive neuroectodermal (BAR 1.00).

CONCLUSION

(1)H MRS has excellent accuracy for the non-invasive diagnosis of common childhood brain tumours particularly if the metabolite information is maximised and should become part of routine clinical assessment for these children.

Robust discrimination of glioblastomas from metastatic brain tumors on the basis of single-voxel (1)H MRS

This article investigates methods for the accurate and robust differentiation of metastases from glioblastomas on the basis of single-voxel (1)H MRS information. Single-voxel (1)H MR spectra from a total of 109 patients (78 glioblastomas and 31 metastases) from the multicenter, international INTERPRET database, plus a test set of 40 patients (30 glioblastomas and 10 metastases) from three different centers in the Barcelona (Spain) metropolitan area, were analyzed using a robust method for feature (spectral frequency) selection coupled with a linear-in-the-parameters single-layer perceptron classifier. For the test set, a parsimonious selection of five frequencies yielded an area under the receiver operating characteristic curve of 0.86, and an area under the convex hull of the receiver operating characteristic curve of 0.91. Moreover, these accurate results for the discrimination between glioblastomas and metastases were obtained using a small number of frequencies that are amenable to metabolic interpretation, which should ease their use as diagnostic markers. Importantly, the prediction can be expressed as a simple formula based on a linear combination of these frequencies. As a result, new cases could be straightforwardly predicted by integrating this formula into a computer-based medical decision support system. This work also shows that the combination of spectra acquired at different TEs (short TE, 20-32 ms; long TE, 135-144 ms) is key to the successful discrimination between glioblastomas and metastases from single-voxel (1)H MRS.

Prospective diagnostic performance evaluation of single-voxel 1H MRS for typing and grading of brain tumours

The purpose of this study was to evaluate whether single-voxel (1)H MRS could add useful information to conventional MRI in the preoperative characterisation of the type and grade of brain tumours. MRI and MRS examinations from a prospective cohort of 40 consecutive patients were analysed double blind by radiologists and spectroscopists before the histological diagnosis was known. The spectroscopists had only the MR spectra, whereas the radiologists had both the MR images and basic clinical details (age, sex and presenting symptoms). Then, the radiologists and spectroscopists exchanged their predictions and re-evaluated their initial opinions, taking into account the new evidence. Spectroscopists used four different systems of analysis for (1)H MRS data, and the efficacy of each of these methods was also evaluated. Information extracted from (1)H MRS significantly improved the radiologists' MRI-based characterisation of grade IV tumours (glioblastomas, metastases, medulloblastomas and lymphomas) in the cohort [area under the curve (AUC) in the MRI re-evaluation 0.93 versus AUC in the MRI evaluation 0.85], and also of the less malignant glial tumours (AUC in the MRI re-evaluation 0.93 versus AUC in the MRI evaluation 0.81). One of the MRS analysis systems used, the INTERPRET (International Network for Pattern Recognition of Tumours Using Magnetic Resonance) decision support system, outperformed the others, as well as being better than the MRI evaluation for the characterisation of grade III astrocytomas. Thus, preoperative MRS data improve the radiologists' performance in diagnosing grade IV tumours and, for those of grade II-III, MRS data help them to recognise the glial lineage. Even in cases in which their diagnoses were not improved, the provision of MRS data to the radiologists had no negative influence on their predictions.

Non-negative matrix factorisation methods for the spectral decomposition of MRS data from human brain tumours

BACKGROUND

In-vivo single voxel proton magnetic resonance spectroscopy (SV 1H-MRS), coupled with supervised pattern recognition (PR) methods, has been widely used in clinical studies of discrimination of brain tumour types and follow-up of patients bearing abnormal brain masses. SV 1H-MRS provides useful biochemical information about the metabolic state of tumours and can be performed at short (< 45 ms) or long (> 45 ms) echo time (TE), each with particular advantages. Short-TE spectra are more adequate for detecting lipids, while the long-TE provides a much flatter signal baseline in between peaks but also negative signals for metabolites such as lactate. Both, lipids and lactate, are respectively indicative of specific metabolic processes taking place. Ideally, the information provided by both TE should be of use for clinical purposes. In this study, we characterise the performance of a range of Non-negative Matrix Factorisation (NMF) methods in two respects: first, to derive sources correlated with the mean spectra of known tissue types (tumours and normal tissue); second, taking the best performing NMF method for source separation, we compare its accuracy for class assignment when using the mixing matrix directly as a basis for classification, as against using the method for dimensionality reduction (DR). For this, we used SV 1H-MRS data with positive and negative peaks, from a widely tested SV 1H-MRS human brain tumour database.

RESULTS

The results reported in this paper reveal the advantage of using a recently described variant of NMF, namely Convex-NMF, as an unsupervised method of source extraction from SV1H-MRS. Most of the sources extracted in our experiments closely correspond to the mean spectra of some of the analysed tumour types. This similarity allows accurate diagnostic predictions to be made both in fully unsupervised mode and using Convex-NMF as a DR step previous to standard supervised classification. The obtained results are comparable to, or more accurate than those obtained with supervised techniques.

CONCLUSIONS

The unsupervised properties of Convex-NMF place this approach one step ahead of classical label-requiring supervised methods for the discrimination of brain tumour types, as it accounts for their increasingly recognised molecular subtype heterogeneity. The application of Convex-NMF in computer assisted decision support systems is expected to facilitate further improvements in the uptake of MRS-derived information by clinicians.

Incremental Gaussian Discriminant Analysis based on Graybill and Deal weighted combination of estimators for brain tumour diagnosis

In the last decade, machine learning (ML) techniques have been used for developing classifiers for automatic brain tumour diagnosis. However, the development of these ML models rely on a unique training set and learning stops once this set has been processed. Training these classifiers requires a representative amount of data, but the gathering, preprocess, and validation of samples is expensive and time-consuming. Therefore, for a classical, non-incremental approach to ML, it is necessary to wait long enough to collect all the required data. In contrast, an incremental learning approach may allow us to build an initial classifier with a smaller number of samples and update it incrementally when new data are collected. In this study, an incremental learning algorithm for Gaussian Discriminant Analysis (iGDA) based on the Graybill and Deal weighted combination of estimators is introduced. Each time a new set of data becomes available, a new estimation is carried out and a combination with a previous estimation is performed. iGDA does not require access to the previously used data and is able to include new classes that were not in the original analysis, thus allowing the customization of the models to the distribution of data at a particular clinical center. An evaluation using five benchmark databases has been used to evaluate the behaviour of the iGDA algorithm in terms of stability-plasticity, class inclusion and order effect. Finally, the iGDA algorithm has been applied to automatic brain tumour classification with magnetic resonance spectroscopy, and compared with two state-of-the-art incremental algorithms. The empirical results obtained show the ability of the algorithm to learn in an incremental fashion, improving the performance of the models when new information is available, and converging in the course of time. Furthermore, the algorithm shows a negligible instance and concept order effect, avoiding the bias that such effects could introduce.

A Web-accessible distributed data warehouse for brain tumour diagnosis

Currently, biological databases (DBs) are a common tool to complement the research of a wide range of biomedical disciplines, but there are only a few specialized medical DBs for human brain tumour magnetic resonance spectroscopy (MRS) data; they typically store a limited range of biological data (i.e. clinical information, magnetic resonance imaging and MRS data) and are not offered as open-source Structured Query Language relational DB schemas. We present a novel approach to biological DBs: a distributed Web-accessible DB for storing and managing clinical and biomedical data related to brain tumours from different clinical centres. This tool is designed for multi-platform systems with dissimilar DB management systems. Being the main data repository of the HealthAgents (HA) project, it uses multi-agent technology and allows the centres to share data and obtain diagnosis classifications from other centres distributed around the world in a reliable way.

The HA project aims to create an agent-based distributed decision support system (DSS) to assist doctors to provide a brain tumour diagnosis and prognosis. The HA DB enables the DSS to totally integrate with its Graphical User Interface to perform classifications with the stored data and visualize the results using the HA distributed agents framework. This new feature converts the system presented in the first application in the world to combine a storage and management tool for brain tumour data and a complete Web-based DSS to obtain automatic diagnosis.

Semi-supervised analysis of human brain tumours from partially labeled MRS information, using manifold learning models

Medical diagnosis can often be understood as a classification problem. In oncology, this typically involves differentiating between tumour types and grades, or some type of discrete outcome prediction. From the viewpoint of computer-based medical decision support, this classification requires the availability of accurate diagnoses of past cases as training target examples. The availability of such labeled databases is scarce in most areas of oncology, and especially so in neuro-oncology. In such context, semi-supervised learning oriented towards classification can be a sensible data modeling choice. In this study, semi-supervised variants of Generative Topographic Mapping, a model of the manifold learning family, are applied to two neuro-oncology problems: the diagnostic discrimination between different brain tumour pathologies, and the prediction of outcomes for a specific type of aggressive brain tumours. Their performance compared favorably with those of the alternative Laplacian Eigenmaps and Semi-Supervised SVM for Manifold Learning models in most of the experiments.

Finding discriminative subtypes of aggressive brain tumours using magnetic resonance spectroscopy

Aggressive tumour types such as glioblastomas (gl) and metastases (me) are known to be difficult to discriminate on the basis of single-voxel proton magnetic resonance spectroscopy (SV 1H-MRS) information. Each of them is also heterogeneous in nature and a statistically robust subtyping analysis is likely to shed light on their structure and, possibly, on their differences. In this brief paper we carry out such analysis. From the original MRS frequencies and their first derivative approximation, the most discriminant variables are first selected by χ(2)-testing. Subtypes are then discovered in the distribution of gl and me by repeated model based cluster analysis. Then, the mean of each subtype is contrasted with the original distribution of MRS spectra by t-testing with tail probabilities for the proportion of false positive (TPPFP) control. Finally, the distribution of gl and me in each subtype is compared with random expectation by χ(2)-testing. The experimental results confirm the existence of consistent subtypes. They exhibit relative proportions of gl and me very unlikely to occur at random.

Brain tumour classification using Gaussian decomposition and neural networks

The development, implementation and use of computer-based medical decision support systems (MDSS) based on pattern recognition techniques holds the promise of substantially improving the quality of medical practice in diagnostic and prognostic tasks. In this study, the core of a decision support system for brain tumour classification from magnetic resonance spectroscopy (MRS) data is presented. It combines data pre-processing using Gaussian decomposition, dimensionality reduction using moving window with variance analysis, and classification using artificial neural networks (ANN). This combination of techniques is shown to yield high diagnostic classification accuracy in problems concerning diverse brain tumour pathologies, some of which have received little attention in the literature.

The INTERPRET Decision-Support System version 3.0 for evaluation of Magnetic Resonance Spectroscopy data from human brain tumours and other abnormal brain masses

BACKGROUND

Proton Magnetic Resonance (MR) Spectroscopy (MRS) is a widely available technique for those clinical centres equipped with MR scanners. Unlike the rest of MR-based techniques, MRS yields not images but spectra of metabolites in the tissues. In pathological situations, the MRS profile changes and this has been particularly described for brain tumours. However, radiologists are frequently not familiar to the interpretation of MRS data and for this reason, the usefulness of decision-support systems (DSS) in MRS data analysis has been explored.

RESULTS

This work presents the INTERPRET DSS version 3.0, analysing the improvements made from its first release in 2002. Version 3.0 is aimed to be a program that 1st, can be easily used with any new case from any MR scanner manufacturer and 2nd, improves the initial analysis capabilities of the first version. The main improvements are an embedded database, user accounts, more diagnostic discrimination capabilities and the possibility to analyse data acquired under additional data acquisition conditions. Other improvements include a customisable graphical user interface (GUI). Most diagnostic problems included have been addressed through a pattern-recognition based approach, in which classifiers based on linear discriminant analysis (LDA) were trained and tested.

CONCLUSIONS

The INTERPRET DSS 3.0 allows radiologists, medical physicists, biochemists or, generally speaking, any person with a minimum knowledge of what an MR spectrum is, to enter their own SV raw data, acquired at 1.5 T, and to analyse them. The system is expected to help in the categorisation of MR Spectra from abnormal brain masses.

SpectraClassifier 1.0: a user friendly, automated MRS-based classifier-development system

BACKGROUND

SpectraClassifier (SC) is a Java solution for designing and implementing Magnetic Resonance Spectroscopy (MRS)-based classifiers. The main goal of SC is to allow users with minimum background knowledge of multivariate statistics to perform a fully automated pattern recognition analysis. SC incorporates feature selection (greedy stepwise approach, either forward or backward), and feature extraction (PCA). Fisher Linear Discriminant Analysis is the method of choice for classification. Classifier evaluation is performed through various methods: display of the confusion matrix of the training and testing datasets; K-fold cross-validation, leave-one-out and bootstrapping as well as Receiver Operating Characteristic (ROC) curves.

RESULTS

SC is composed of the following modules: Classifier design, Data exploration, Data visualisation, Classifier evaluation, Reports, and Classifier history. It is able to read low resolution in-vivo MRS (single-voxel and multi-voxel) and high resolution tissue MRS (HRMAS), processed with existing tools (jMRUI, INTERPRET, 3DiCSI or TopSpin). In addition, to facilitate exchanging data between applications, a standard format capable of storing all the information needed for a dataset was developed. Each functionality of SC has been specifically validated with real data with the purpose of bug-testing and methods validation. Data from the INTERPRET project was used.

CONCLUSIONS

SC is a user-friendly software designed to fulfil the needs of potential users in the MRS community. It accepts all kinds of pre-processed MRS data types and classifies them semi-automatically, allowing spectroscopists to concentrate on interpretation of results with the use of its visualisation tools.

Diagnosis of brain tumours from magnetic resonance spectroscopy using wavelets and Neural Networks

The diagnosis of human brain tumours from noninvasive signal measurements is a sensitive task that requires specialized expertise. In this task, radiology experts are likely to benefit from the support of computer-based systems built around robust classification processes. In this brief paper, a method that combines data pre-processing using wavelets with classification using Artificial Neural Networks is shown to yield high diagnostic classification accuracy for a broad range of brain tumour pathologies.

Multiproject-multicenter evaluation of automatic brain tumor classification by magnetic resonance spectroscopy

JUSTIFICATION

Automatic brain tumor classification by MRS has been under development for more than a decade. Nonetheless, to our knowledge, there are no published evaluations of predictive models with unseen cases that are subsequently acquired in different centers. The multicenter eTUMOUR project (2004-2009), which builds upon previous expertise from the INTERPRET project (2000-2002) has allowed such an evaluation to take place.

MATERIALS AND METHODS

A total of 253 pairwise classifiers for glioblastoma, meningioma, metastasis, and low-grade glial diagnosis were inferred based on 211 SV short TE INTERPRET MR spectra obtained at 1.5 T (PRESS or STEAM, 20-32 ms) and automatically pre-processed. Afterwards, the classifiers were tested with 97 spectra, which were subsequently compiled during eTUMOUR.

RESULTS

In our results based on subsequently acquired spectra, accuracies of around 90% were achieved for most of the pairwise discrimination problems. The exception was for the glioblastoma versus metastasis discrimination, which was below 78%. A more clear definition of metastases may be obtained by other approaches, such as MRSI + MRI.

CONCLUSIONS

The prediction of the tumor type of in-vivo MRS is possible using classifiers developed from previously acquired data, in different hospitals with different instrumentation under the same acquisition protocols. This methodology may find application for assisting in the diagnosis of new brain tumor cases and for the quality control of multicenter MRS databases.

The effect of combining two echo times in automatic brain tumor classification by MRS

(1)H MRS is becoming an accurate, non-invasive technique for initial examination of brain masses. We investigated if the combination of single-voxel (1)H MRS at 1.5 T at two different (TEs), short TE (PRESS or STEAM, 20-32 ms) and long TE (PRESS, 135-136 ms), improves the classification of brain tumors over using only one echo TE. A clinically validated dataset of 50 low-grade meningiomas, 105 aggressive tumors (glioblastoma and metastasis), and 30 low-grade glial tumors (astrocytomas grade II, oligodendrogliomas and oligoastrocytomas) was used to fit predictive models based on the combination of features from short-TEs and long-TE spectra. A new approach that combines the two consecutively was used to produce a single data vector from which relevant features of the two TE spectra could be extracted by means of three algorithms: stepwise, reliefF, and principal components analysis. Least squares support vector machines and linear discriminant analysis were applied to fit the pairwise and multiclass classifiers, respectively. Significant differences in performance were found when short-TE, long-TE or both spectra combined were used as input. In our dataset, to discriminate meningiomas, the combination of the two TE acquisitions produced optimal performance. To discriminate aggressive tumors from low-grade glial tumours, the use of short-TE acquisition alone was preferable. The classifier development strategy used here lends itself to automated learning and test performance processes, which may be of use for future web-based multicentric classifier development studies.

MRS quality assessment in a multicentre study on MRS-based classification of brain tumours

This paper reports on quality assessment of MRS in the European Union-funded multicentre project INTERPRET (International Network for Pattern Recognition of Tumours Using Magnetic Resonance; http://azizu.uab.es/INTERPRET), which has developed brain tumour classification software using in vivo proton MR spectra. The quality assessment consisted of both MR system quality assurance (SQA) and quality control (QC) of spectral data acquired from patients and healthy volunteers. The system performance of the MR spectrometers at all participating centres was checked bimonthly by a short measurement protocol using a specially designed INTERPRET phantom. In addition, a more extended SQA protocol was performed yearly and after each hardware or software upgrade. To compare the system performance for in vivo measurements, each centre acquired MR spectra from the brain of five healthy volunteers. All MR systems fulfilled generally accepted minimal system performance for brain MRS during the entire data acquisition period. The QC procedure of the MR spectra in the database comprised automatic determination of the signal-to-noise ratio (SNR) in a water-suppressed spectrum and of the line width of the water resonance (water band width, WBW) in the corresponding non-suppressed spectrum. Values of SNR > 10 and WBW < 8 Hz at 1.5 T were determined empirically as conservative threshold levels required for spectra to be of acceptable quality. These thresholds only hold for SNR and WBW values using the definitions and data processing described in this article. A final QC check consisted of visual inspection of each clinically validated water-suppressed metabolite spectrum by two, or, in the case of disagreement, three, experienced MR spectroscopists, to detect artefacts such as large baseline distortions, exceptionally broadened metabolite peaks, insufficient removal of the water line, large phase errors, and signals originating from outside the voxel. In the end, 10% of 889 spectra with completed spectroscopic judgement were discarded.

Applications of magnetic resonance spectroscopy in radiotherapy treatment planning

Following advances in conformal radiotherapy, a key problem now facing radiation oncologists is target definition. While MRI and CT provide images of excellent spatial resolution, they do not always provide sufficient contrast to identify tumour extent or to identify regions of high cellular activity that might be targeted with boost doses. Magnetic resonance spectroscopy (MRS) is an alternative approach that holds great promise for aiding target definition for radiotherapy treatment planning, and for evaluation of response and recurrence. MRS is able to detect signals from low molecular weight metabolites such as choline and creatine that are present at concentrations of a few mM in tissue. Spectra may be acquired from single voxels, or from a 2D or 3D array of voxels using spectroscopic imaging. The current state of the art achieves a spatial resolution of 6-10 mm in a scan time of about 10-15 min. Co-registered MR images are acquired in the same examination. The method is currently under evaluation, in particular in brain (where MRS has been shown to differentiate between many tumour types and grades) and in prostate (where cancer may be distinguished from normal tissue and benign prostatic hypertrophy). The contrast achieved with MRS, based on tissue biochemistry, therefore provides a promising alternative for identifying tumour extent and regions of high metabolic activity. It is anticipated that MRS will become an essential tool for treatment planning where other modalities lack the necessary contrast.

Magnetic resonance spectroscopy (MRS) in the investigation of cancer at The Royal Marsden Hospital and The Institute of Cancer Research

Developments in magnetic resonance spectroscopy (MRS) at The Royal Marsden Hospital and The Institute of Cancer Research are reviewed in the context of preceding developments in nuclear magnetic resonance (NMR) and MRS, and some of the early developments in this field, particularly those leading to human measurements. The early development of technology, and associated techniques for human measurement and assessment will be discussed, with particular reference to experience at out institutions. Applications using particular nuclei will then be described and related to other experimental work where appropriate. Contributions to the development of MRS that have been published in Physics in Medicine and Biology will be discussed.