NIMG-36. VISUALIZATION OF TUMOR HETEROGENEITY AND PREDICTION OF ISOCITRATE DEHYDROGENASE MUTATION STATUS FOR HUMAN GLIOMAS BY USING MULTIPARAMETRIC PHYSIOLOGIC AND METABOLIC MRI

2021 ◽  
Vol 23 (Supplement_6) ◽  
pp. vi136-vi137
Author(s):  
Akifumi Hagiwara ◽  
Hiroyuki Tatekawa ◽  
Yao Jingwen ◽  
Catalina Raymond ◽  
Richard Everson ◽  
...  
Keyword(s):  
Machine Learning ◽  
Sensitivity And Specificity ◽  
Isocitrate Dehydrogenase ◽  
Glucose Transporter ◽  
Support Vector ◽  
Diffuse Glioma ◽  
Idh Mutation ◽  
Self Organizing Map ◽  
Mutation Status ◽  
Isocitrate Dehydrogenase Mutation

Abstract Preoperative prediction of isocitrate dehydrogenase mutation status is clinically meaningful, but remains challenging. This study aimed to predict the isocitrate dehydrogenase (IDH) status of gliomas by using the machine learning voxel-wise clustering method of multiparametric physiologic and metabolic magnetic resonance imaging (MRI) and to show the association of the created cluster labels with the glucose metabolism status of the tumors. Sixty-nine patients with diffuse glioma were scanned by pH-sensitive MRI, diffusion-weighted imaging, fluid-attenuated inversion recovery, and contrast-enhanced T1-weighted imaging at 3 T. An unsupervised two-level clustering approach, including the generation of a self-organizing map followed by the K-means clustering, was used for voxel-wise feature extraction from the acquired images. The logarithmic ratio of the labels in each class within tumor regions was applied to a support vector machine to differentiate IDH mutation status. Bootstrapping and leave-one-out cross-validation were used to calculate the area under the curve (AUC) of receiver operating characteristic curves, accuracy, sensitivity, and specificity for evaluating performance. Targeted biopsies were performed for 14 patients to explore the relationship between clustered labels and the expression of key glycolytic proteins determined using immunohistochemistry. The highest prediction performance to differentiate IDH status was found for 10-class clustering, with a mean AUC, accuracy, sensitivity, and specificity of 0.94, 0.91, 0.90, and 0.91, respectively. The tissues with labels 7 + 8 + 9 + 10 showed high expression levels of hypoxia-inducible factor 1-alpha, glucose transporter 3, and hexokinase 2, which are typical of IDH wild-type glioma, whereas those with labels 1 showed low expression of these proteins. Our machine learning model successfully predicted the IDH mutation status of gliomas, and the resulting clusters properly reflected the metabolic status of the tumors.

scholarly journals Differentiating IDH status in human gliomas using machine learning and multiparametric MR/PET

2021 ◽  
Vol 21 (1) ◽  
Author(s):  
Hiroyuki Tatekawa ◽  
Akifumi Hagiwara ◽  
Hiroyuki Uetani ◽  
Shadfar Bahri ◽  
Catalina Raymond ◽  
...  
Keyword(s):  
Machine Learning ◽  
Support Vector Machine ◽  
Cerebral Blood Volume ◽  
Area Under The Curve ◽  
Support Vector ◽  
Idh Mutation ◽  
Self Organizing Map ◽  
Mutation Status ◽  
Multiparametric Imaging ◽  
Clustering Approach

Abstract Background The purpose of this study was to develop a voxel-wise clustering method of multiparametric magnetic resonance imaging (MRI) and 3,4-dihydroxy-6-[18F]-fluoro-L-phenylalanine (FDOPA) positron emission tomography (PET) images using an unsupervised, two-level clustering approach followed by support vector machine in order to classify the isocitrate dehydrogenase (IDH) status of gliomas. Methods Sixty-two treatment-naïve glioma patients who underwent FDOPA PET and MRI were retrospectively included. Contrast enhanced T1-weighted images, T2-weighted images, fluid-attenuated inversion recovery images, apparent diffusion coefficient maps, and relative cerebral blood volume maps, and FDOPA PET images were used for voxel-wise feature extraction. An unsupervised two-level clustering approach, including a self-organizing map followed by the K-means algorithm was used, and each class label was applied to the original images. The logarithmic ratio of labels in each class within tumor regions was applied to a support vector machine to differentiate IDH mutation status. The area under the curve (AUC) of receiver operating characteristic curves, accuracy, and F1-socore were calculated and used as metrics for performance. Results The associations of multiparametric imaging values in each cluster were successfully visualized. Multiparametric images with 16-class clustering revealed the highest classification performance to differentiate IDH status with the AUC, accuracy, and F1-score of 0.81, 0.76, and 0.76, respectively. Conclusions Machine learning using an unsupervised two-level clustering approach followed by a support vector machine classified the IDH mutation status of gliomas, and visualized voxel-wise features from multiparametric MRI and FDOPA PET images. Unsupervised clustered features may improve the understanding of prioritizing multiparametric imaging for classifying IDH status.

Intraoperative assessment of isocitrate dehydrogenase mutation status in human gliomas using desorption electrospray ionization–mass spectrometry

2020 ◽  
Vol 132 (1) ◽  
pp. 180-187 ◽  
Author(s):  
Clint M. Alfaro ◽  
Valentina Pirro ◽  
Michael F. Keating ◽  
Eyas M. Hattab ◽  
R. Graham Cooks ◽  
...  
Keyword(s):  
Mass Spectrometry ◽  
Electrospray Ionization ◽  
Isocitrate Dehydrogenase ◽  
Tumor Resection ◽  
Desorption Electrospray Ionization ◽  
Extent Of Resection ◽  
Ionization Mass ◽  
Idh Mutation ◽  
Mutation Status ◽  
Desi Ms

OBJECTIVEThe authors describe a rapid intraoperative ambient ionization mass spectrometry (MS) method for determining isocitrate dehydrogenase (IDH) mutation status from glioma tissue biopsies. This method offers new glioma management options and may impact extent of resection goals. Assessment of the IDH mutation is key for accurate glioma diagnosis, particularly for differentiating diffuse glioma from other neoplastic and reactive inflammatory conditions, a challenge for the standard intraoperative diagnostic consultation that relies solely on morphology.METHODSBanked glioma specimens (n = 37) were analyzed by desorption electrospray ionization–MS (DESI-MS) to develop a diagnostic method to detect the known altered oncometabolite in IDH-mutant gliomas, 2-hydroxyglutarate (2HG). The method was used intraoperatively to analyze tissue smears obtained from glioma patients undergoing resection and to rapidly diagnose IDH mutation status (< 5 minutes). Fifty-one tumor core biopsies from 25 patients (14 wild type [WT] and 11 mutant) were examined and data were analyzed using analysis of variance and receiver operating characteristic curve analysis.RESULTSThe optimized DESI-MS method discriminated between IDH-WT and IDH-mutant gliomas, with an average sensitivity and specificity of 100%. The average normalized DESI-MS 2HG signal was an order of magnitude higher in IDH-mutant glioma than in IDH-WT glioma. The DESI 2HG signal intensities correlated with independently measured 2HG concentrations (R2 = 0.98). In 1 case, an IDH1 R132H–mutant glioma was misdiagnosed as a demyelinating condition by frozen section histology during the intraoperative consultation, and no resection was performed pending the final pathology report. A second craniotomy and tumor resection was performed after the final pathology provided a diagnosis most consistent with an IDH-mutant glioblastoma. During the second craniotomy, high levels of 2HG in the tumor core biopsies were detected.CONCLUSIONSThis study demonstrates the capability to differentiate rapidly between IDH-mutant gliomas and IDH-WT conditions by DESI-MS during tumor resection. DESI-MS analysis of tissue smears is simple and can be easily integrated into the standard intraoperative pathology consultation. This approach may aid in solving differential diagnosis problems associated with low-grade gliomas and could influence intraoperative decisions regarding extent of resection, ultimately improving patient outcome. Research is ongoing to expand the patient cohort, systematically validate the DESI-MS method, and investigate the relationships between 2HG and tumor heterogeneity.

scholarly journals T157. THE COURSE OF SCHIZOPHRENIA-RELATED NEURAL FINGERPRINTS OVER NINE YEARS - A LONGITUDINAL POPULATION-BASED MACHINE LEARNING STUDY

2020 ◽  
Vol 46 (Supplement_1) ◽  
pp. S290-S291
Author(s):  
Johannes Lieslehto ◽  
Erika Jääskeläinen ◽  
Jouko Miettunen ◽  
Matti Isohanni ◽  
Dominic Dwyer ◽  
...  
Keyword(s):  
Machine Learning ◽  
Sensitivity And Specificity ◽  
Verbal Learning ◽  
Clinical Variable ◽  
Support Vector ◽  
Cohen’S D ◽  
Low Degree ◽  
Cohen's D ◽  
Over Time

Abstract Background Previous machine learning studies using structural MRI (sMRI) have been able to separate schizophrenia from controls with relatively high (about 80%) sensitivity and specificity (Kambeitz et al. Neuropsychopharmacology 2015). Interestingly, prediction accuracy in first-episode psychosis is lower compared to older and probably more chronic patients. One possibility is that the appearance of the neurodiagnostic fingerprints (NF) originated from the schizophrenia vs. controls classifier become more visible over time in schizophrenia due to the progressive nature of the disorder. Methods Using the Cobre sample (70 schizophrenia and 74 controls), we trained support vector machine (SVM) to differentiate schizophrenia from controls using sMRI. Next, we utilized the Northern Finland Birth Cohort 1966 (NFBC 1966) sample of 29 schizophrenia and 61 non-psychotic controls who participated in the nine-year follow-up. We applied the Cobre-trained SVM models at the baseline (participants 34 years old) and the follow-up (participants 43 years old) using out of sample cross-validation without any in-between retraining. Two independent schizophrenia datasets (the Neuromorphometry by Computer Algorithm Chicago [NMorphCH] and the Consortium for Neuropsychiatric Phenomics [CNP]) were utilized for replication analyses of the SVM generalizability. To address the possibility that the NF mainly capture some general psychopathology, we tested whether the NF generalize to depression using two independent MDD samples from Munich and Münster, Germany. Results Using the Cobre-trained SVM models for schizophrenia vs. controls differentiation in the NFBC 1966, we found balanced accuracy (i.e. mean of sensitivity and specificity, [BAC]) of 72.8% (sensitivity=58.6%, specificity=86.9%) at the baseline and BAC of 79.7% (sensitivity=75.9%, specificity=83.6%) at the follow-up. In the NFBC 1966 schizophrenia patients, we found that SVM decision scores varied as a function of timepoint into the direction of more schizophrenia-likeness at the follow-up (paired T-test, Cohen’s d=0.58, P=0.004). The same was not true in controls (Cohen’s d=0.09, P=0.49). The SVM decision score difference*timepoint interaction related to the decrease of hippocampus and medial prefrontal cortex. The SVM models’ performance was also validated at the two replication samples (BAC of 77.5% in the CNP and BAC of 69.1% in the NMorphCH). In the NFBC 1966 the strongest clinical variable correlating with the trajectory of SVM decision scores over the follow-up was poor performance in the California Verbal Learning Test. This finding was also replicated in the CNP dataset. Further, in the NFBC 1966, those schizophrenia patients with a low degree of SVM decision scores had a higher probability of being in remission, being able to work, and being without antipsychotic medication at the follow-up. The generalization of the SVM models to MDD was worse compared to schizophrenia classification (DeLong’s tests for the two ROC curves: P&lt;0.001). Discussion The degree of schizophrenia-related neurodiagnostic fingerprints appear to magnify over time in schizophrenia. By contrast, the discernibility of these fingerprints in controls does not change over time. This indicates that the NF captures some schizophrenia-related progressive neural changes, and not, e.g., normal aging-related brain volume loss. The fingerprints were also generalizable to other schizophrenia samples. Further, the fingerprints seem to have some disorder specificity as the SVM models do not generalize to depression. Lastly, it appears that a low degree of schizophrenia-related NF in schizophrenia might possess some value in predicting patients’ future remission and recovery-related factors.

scholarly journals Non-Invasive Prediction of IDH Mutation in Patients with Glioma WHO II/III/IV Based on F-18-FET PET-Guided In Vivo 1H-Magnetic Resonance Spectroscopy and Machine Learning

Cancers ◽  
2020 ◽  
Vol 12 (11) ◽  
pp. 3406
Author(s):  
Elisabeth Bumes ◽  
Fro-Philip Wirtz ◽  
Claudia Fellner ◽  
Jirka Grosse ◽  
Dirk Hellwig ◽  
...  
Keyword(s):  
Machine Learning ◽  
Magnetic Resonance ◽  
Magnetic Resonance Spectroscopy ◽  
Support Vector ◽  
Resonance Spectroscopy ◽  
Idh Mutation ◽  
1H Mrs ◽  
1H Magnetic Resonance Spectroscopy ◽  
Non Invasive

Isocitrate dehydrogenase (IDH)-1 mutation is an important prognostic factor and a potential therapeutic target in glioma. Immunohistological and molecular diagnosis of IDH mutation status is invasive. To avoid tumor biopsy, dedicated spectroscopic techniques have been proposed to detect D-2-hydroxyglutarate (2-HG), the main metabolite of IDH, directly in vivo. However, these methods are technically challenging and not broadly available. Therefore, we explored the use of machine learning for the non-invasive, inexpensive and fast diagnosis of IDH status in standard 1H-magnetic resonance spectroscopy (1H-MRS). To this end, 30 of 34 consecutive patients with known or suspected glioma WHO grade II-IV were subjected to metabolic positron emission tomography (PET) imaging with O-(2-18F-fluoroethyl)-L-tyrosine (18F-FET) for optimized voxel placement in 1H-MRS. Routine 1H-magnetic resonance (1H-MR) spectra of tumor and contralateral healthy brain regions were acquired on a 3 Tesla magnetic resonance (3T-MR) scanner, prior to surgical tumor resection and molecular analysis of IDH status. Since 2-HG spectral signals were too overlapped for reliable discrimination of IDH mutated (IDHmut) and IDH wild-type (IDHwt) glioma, we used a nested cross-validation approach, whereby we trained a linear support vector machine (SVM) on the complete spectral information of the 1H-MRS data to predict IDH status. Using this approach, we predicted IDH status with an accuracy of 88.2%, a sensitivity of 95.5% (95% CI, 77.2–99.9%) and a specificity of 75.0% (95% CI, 42.9–94.5%), respectively. The area under the curve (AUC) amounted to 0.83. Subsequent ex vivo 1H-nuclear magnetic resonance (1H-NMR) measurements performed on metabolite extracts of resected tumor material (eight specimens) revealed myo-inositol (M-ins) and glycine (Gly) to be the major discriminators of IDH status. We conclude that our approach allows a reliable, non-invasive, fast and cost-effective prediction of IDH status in a standard clinical setting.

Quantitative Assessment of Ataxia in Multiple Sclerosis Patients using Spatiotemporal Parameters: A Relief-Based Machine Learning Analysis

2021 ◽  
Author(s):  
Furkan Bilek ◽  
Ferhat Balgetir ◽  
Caner Feyzi Demir ◽  
Gökhan Alkan ◽  
Seda Arslan-Tuncer
Keyword(s):  
Machine Learning ◽  
Multiple Sclerosis ◽  
Sensitivity And Specificity ◽  
Classification Performance ◽  
Support Vector ◽  
Learning Methods ◽  
K Nearest Neighbors ◽  
Machine Learning Methods ◽  
Spatiotemporal Parameters ◽  
Multiple Sclerosis Patients

Abstract Background and Objective Multiple sclerosis (MS) is a chronic, progressive, and autoimmune disease of the central nervous system (CNS) characterized by inflammation, demyelination, and axonal injury. In patients with newly diagnosed MS (ndMS), ataxia can present either as mild or severe and can be difficult to diagnose in the absence of clinical disability. Such difficulties can be eliminated by using decision support systems supported by machine learning methods. The present study aimed to achieve early diagnosis of ataxia in ndMS patients by using machine learning methods with spatiotemporal parameters. Materials and Methods The prospective study included 32 ndMS patients with an Expanded Disability Status Scale (EDSS) score of≤2.0 and 32 healthy volunteers. A total of 14 parameters were elicited by using a Win-Track platform. The ndMS patients were differentiated from healthy individuals using multiple classifiers including Artificial Neural Network (ANN), Support Vector Machine (SVM), the k-nearest neighbors (K-NN) algorithm, and Decision Tree Learning (DTL). To improve the performance of the classification, a Relief-based feature selection algorithm was applied to select the subset that best represented the whole dataset. Performance evaluation was achieved based on several criteria such as Accuracy (ACC), Sensitivity (SN), Specificity (SP), and Precision (PREC). Results ANN had a higher classification performance compared to other classifiers, whereby it provided an accuracy, sensitivity, and specificity of 89, 87.8, 90.3% with the use of all parameters and provided the values of 93.7, 96.6%, and 91.1% with the use of parameters selected by the Relief algorithm, respectively. Significance To our knowledge, this is the first study of its kind in the literature to investigate the diagnosis of ataxia in ndMS patients by using machine learning methods with spatiotemporal parameters. The proposed method, i. e. Relief-based ANN method, successfully diagnosed ataxia by using a lower number of parameters compared to the numbers of parameters reported in clinical studies, thereby reducing the costs and increasing the performance of the diagnosis. The method also provided higher rates of accuracy, sensitivity, and specificity in the diagnosis of ataxia in ndMS patients compared to other methods. Taken together, these findings indicate that the proposed method could be helpful in the diagnosis of ataxia in minimally impaired ndMS patients and could be a pathfinder for future studies.

scholarly journals P01.075 Understanding cognitive functioning in diffuse glioma patients: the relevance of IDH mutation status and functional connectivity

2018 ◽  
Vol 20 (suppl_3) ◽  
pp. iii247-iii247 ◽  
Author(s):  
J Derks ◽  
S Kulik ◽  
P Wesseling ◽  
T Numan ◽  
A Hillebrand ◽  
...  
Keyword(s):  
Functional Connectivity ◽  
Cognitive Functioning ◽  
Diffuse Glioma ◽  
Idh Mutation ◽  
Mutation Status

scholarly journals Aligning text mining and machine learning algorithms with best practices for study selection in systematic literature reviews

2020 ◽  
Vol 9 (1) ◽  
Author(s):  
E. Popoff ◽  
M. Besada ◽  
J. P. Jansen ◽  
S. Cope ◽  
S. Kanters
Keyword(s):  
Machine Learning ◽  
Text Mining ◽  
Sensitivity And Specificity ◽  
Full Text ◽  
Learning Algorithms ◽  
Machine Learning Algorithms ◽  
Training Data ◽  
Support Vector ◽  
Study Selection ◽  
Literature Reviews

Abstract Background Despite existing research on text mining and machine learning for title and abstract screening, the role of machine learning within systematic literature reviews (SLRs) for health technology assessment (HTA) remains unclear given lack of extensive testing and of guidance from HTA agencies. We sought to address two knowledge gaps: to extend ML algorithms to provide a reason for exclusion—to align with current practices—and to determine optimal parameter settings for feature-set generation and ML algorithms. Methods We used abstract and full-text selection data from five large SLRs (n = 3089 to 12,769 abstracts) across a variety of disease areas. Each SLR was split into training and test sets. We developed a multi-step algorithm to categorize each citation into the following categories: included; excluded for each PICOS criterion; or unclassified. We used a bag-of-words approach for feature-set generation and compared machine learning algorithms using support vector machines (SVMs), naïve Bayes (NB), and bagged classification and regression trees (CART) for classification. We also compared alternative training set strategies: using full data versus downsampling (i.e., reducing excludes to balance includes/excludes because machine learning algorithms perform better with balanced data), and using inclusion/exclusion decisions from abstract versus full-text screening. Performance comparisons were in terms of specificity, sensitivity, accuracy, and matching the reason for exclusion. Results The best-fitting model (optimized sensitivity and specificity) was based on the SVM algorithm using training data based on full-text decisions, downsampling, and excluding words occurring fewer than five times. The sensitivity and specificity of this model ranged from 94 to 100%, and 54 to 89%, respectively, across the five SLRs. On average, 75% of excluded citations were excluded with a reason and 83% of these citations matched the reviewers’ original reason for exclusion. Sensitivity significantly improved when both downsampling and abstract decisions were used. Conclusions ML algorithms can improve the efficiency of the SLR process and the proposed algorithms could reduce the workload of a second reviewer by identifying exclusions with a relevant PICOS reason, thus aligning with HTA guidance. Downsampling can be used to improve study selection, and improvements using full-text exclusions have implications for a learn-as-you-go approach.

scholarly journals Machine learning-based CT radiomics model distinguishes COVID-19 from non-COVID-19 pneumonia

2021 ◽  
Vol 21 (1) ◽  
Author(s):  
Hui Juan Chen ◽  
Li Mao ◽  
Yang Chen ◽  
Li Yuan ◽  
Fei Wang ◽  
...  
Keyword(s):  
Machine Learning ◽  
Clinical Features ◽  
Sensitivity And Specificity ◽  
Characteristic Curve ◽  
Automatic Segmentation ◽  
Integrated Model ◽  
Support Vector ◽  
Radiological Features ◽  
Volume Of Interest ◽  
Svm Model

Abstract Background To develop a machine learning-based CT radiomics model is critical for the accurate diagnosis of the rapid spreading coronavirus disease 2019 (COVID-19). Methods In this retrospective study, a total of 326 chest CT exams from 134 patients (63 confirmed COVID-19 patients and 71 non-COVID-19 patients) were collected from January 20 to February 8, 2020. A semi-automatic segmentation procedure was used to delineate the volume of interest (VOI), and radiomic features were extracted. The Support Vector Machine (SVM) model was built on the combination of 4 groups of features, including radiomic features, traditional radiological features, quantifying features, and clinical features. By repeating cross-validation procedure, the performance on the time-independent testing cohort was evaluated by the area under the receiver operating characteristic curve (AUC), accuracy, sensitivity, and specificity. Results For the SVM model built on the combination of 4 groups of features (integrated model), the per-exam AUC was 0.925 (95% CI 0.856 to 0.994) for differentiating COVID-19 on the testing cohort, and the sensitivity and specificity were 0.816 (95% CI 0.651 to 0.917) and 0.923 (95% CI 0.621 to 0.996), respectively. As for the SVM models built on radiomic features, radiological features, quantifying features, and clinical features, individually, the AUC on the testing cohort reached 0.765, 0.818, 0.607, and 0.739, respectively, significantly lower than the integrated model, except for the radiomic model. Conclusion The machine learning-based CT radiomics models may accurately classify COVID-19, helping clinicians and radiologists to identify COVID-19 positive cases.

scholarly journals Machine learning assisted DSC-MRI radiomics as a tool for glioma classification by grade and mutation status

2019 ◽  
Author(s):  
Carole H. Sudre ◽  
Jasmina Panovska-Griffiths ◽  
Eser Sanverdi ◽  
Sebastian Brandner ◽  
Vasileios K. Katsaros ◽  
...  
Keyword(s):  
Machine Learning ◽  
Diagnostic Value ◽  
Dynamic Susceptibility ◽  
Idh Mutation ◽  
Dynamic Susceptibility Contrast ◽  
Mutation Status ◽  
Dsc Mri ◽  
Grade Difference ◽  
Grade Ii ◽  
Susceptibility Contrast

AbstractBackgroundMachine learning assisted MRI radiomics, which combines MRI techniques with machine learning methodology, is rapidly gaining attention as a promising method for staging of brain gliomas. This study assesses the diagnostic value of such framework applied to dynamic susceptibility contrast (DSC)-MRI in classifying treatment-naïve gliomas from a multi-center patient pool into WHO grades II-IV and across their isocitrate dehydrogenase (IDH) mutation status.Methods333 patients from 6 tertiary centres, diagnosed histologically and molecularly with primary gliomas (IDH-mutant=151 or IDH-wildtype=182) were retrospectively identified. Raw DSC-MRI data was post-processed for normalised leakage-corrected relative cerebral blood volume (rCBV) maps. Shape, intensity distribution (histogram) and rotational invariant Haralick texture features over the tumour mask were extracted. Differences in extracted features between IDH-wildtype and IDH-mutant gliomas and across three glioma grades were tested using the Wilcoxon two-sample test. A random forest algorithm was employed (2-fold cross-validation, 250 repeats) to predict grades or mutation status using the extracted features.ResultsFeatures from all types (shape, distribution, texture) showed significant differences across mutation status. WHO grade II-III differentiation was mostly driven by shape features while texture and intensity feature were more relevant for the III-IV separation. Increased number of features became significant when differentiating grades further apart from one another. Gliomas were correctly stratified by IDH mutation status in 71% of the cases and by grade in 53% of the cases. In addition, 87% of the gliomas grades predicted with an error distance up to 1.ConclusionDespite large heterogeneity in the multi-center dataset, machine learning assisted DSC-MRI radiomics hold potential to address the inherent variability and presents a promising approach for non-invasive glioma molecular subtyping and grading.Key points-On highly heterogenous, multi-centre data, machine learning on DSC-MRI features can correctly predict glioma IDH subtyping in 71% of cases and glioma grade II-IV in 53% of the cases (87% <1 grade difference)-Shape features distinguish best grade II from grade III gliomas.-Texture and distribution features distinguish best grade III from grade IV tumours.Importance of studyThis work illustrates the diagnostic value of combining machine learning and dynamic susceptibility contrast-enhanced MRI (DSC-MRI) radiomics in classifying gliomas into WHO grades II-IV as well as across their isocitrate dehydrogenase (IDH) mutation status. Despite the data heterogeneity inherent to the multi-centre design of the studied cohort (333 subjects, 6 centres) that greatly increases the theoretical challenges of machine learning frameworks, good classification performance (accuracy of 53% across grades (87% <1 grade difference) and 71% across mutation status) was obtained. Therefore, our results provide a proof-of-concept for this emerging precision medicine field that has good generalisability and scalability properties. Introspection on the classification errors highlighted mostly borderline cases and helped underline the challenges of a categorical classification in a pathological continuum.With its strong generalisability property, its ability to further incorporate participating centres and its possible use to identify borderline cases, the proposed machine learning framework has the potential to contribute to the clinical translation of machine-learning assisted diagnostic tools in neuro-oncology.

Sign in / Sign up

Export Citation Format

Share Document