Search

Gliomas are the most common primary malignant brain tumors in adults. Accurate grading is crucial as therapeutic strategies are often disparate for different grades and may influence patient prognosis. This study aims to provide an automated glioma grading platform on the basis of machine learning models. In this paper, we investigate contributions of multi-parameters from multimodal data including imaging parameters or features from the Whole Slide images (WSI) and the proliferation marker Ki-67 for automated brain tumor grading. For each WSI, we extract both visual parameters such as morphology parameters and sub-visual parameters including first-order and second-order features. On the basis of machine learning models, our platform classifies gliomas into grades II, III, and IV. Furthermore, we quantitatively interpret and reveal the important parameters contributing to grading with the Local Interpretable Model-Agnostic Explanations (LIME) algorithm. The quantitative analysis and explanation may assist clinicians to better understand the disease and accordingly to choose optimal treatments for improving clinical outcomes. The performance of our grading model was evaluated with cross-validation, which randomly divided the patients into non-overlapping training and testing sets and repeatedly validated the model on the different testing sets. The primary results indicated that this modular platform approach achieved the highest grading accuracy of 0.90 ± 0.04 with support vector machine (SVM) algorithm, with grading accuracies of 0.91 ± 0.08, 0.90 ± 0.08, and 0.90 ± 0.07 for grade II, III, and IV gliomas, respectively.

Background: Accurate glioma grading plays an important role in patient treatment.Purpose: To investigate the influence of varied texture retrieving models on the efficacy of grading glioma with support vector machine (SVM).Study Type: Retrospective.Population: In all, 117 glioma patients including 25, 29, and 63 grade II, III, and IV gliomas, respectively, based on WHO 2007.Field Strength/sequence: 3.0T MRI/ T1 WI, T2 fluid-attenuated inversion recovery, contrast enhanced T1 , arterial spinal labeling, diffusion-weighted imaging (0, 30, 50, 100, 200, 300, 500, 800, 1000, 1500, 2000, 3000, and 3500 sec/mm2 ), and dynamic contrast-enhanced.Assessment: Texture attributes from 30 parametric maps were retrieved using four models, including Global, gray-level co-occurrence matrix (GLCM), gray-level run-length matrix (GLRLM), and gray-level size-zone matrix (GLSZM). Attributes derived from varied models were input into radial basis function SVM (RBF-SVM) combined with attribute selection using SVM-recursive feature elimination (SVM-RFE). The SVM model was trained and established with 80% randomly selected data of each category using 10-fold crossvalidation. The model performance was further tested using the remaining 20% data.Statistical Tests: Ten-fold crossvalidation was used to validate the model performance.Results: Based on 30 parametric maps, 90, 240, 390, or 390 texture attributes were retrieved using the Global, GLCM, GLRLM, or GLSZM model, respectively. SVM-RFE was able to reduce attribute redundancy as well as improve RBF-SVM performance. Training data were oversampled by applying the Synthetic Minority Oversampling Technique (SMOTE) method to overcome the data imbalance problem; test results were able to further demonstrate the classifying performance of the final models. GLSZM using gray-level 64 was the optimal model to retrieve powerful image texture attributes to produce enough classifying power with an accuracy / area under the curve of 0.760/0.867 for the training and 0.875/0.971 for the independent test. Fifteen attributes were selected with SVM-RFE to provide comparable classifying efficacy.Data Conclusion: When using image textures-based SVM classification of gliomas, the GLSZM model in combination with gray-level 64 and attribute selection may be an optimized solution.Level Of Evidence: 2 Technical Efficacy: Stage 2 J. Magn. Reson. Imaging 2019;49:1263-1274. [ABSTRACT FROM AUTHOR]

• Preoperative differentiation of Low vs High glioma grade is of major importance, for precise selection of therapeutic plan. • Radiomic analysis of advanced multiparametric MRI data significantly contributes in the differential diagnosis. • Diffusion, Perfusion and Spectroscopy MRI data provide additional prognostic biomarkers for tumor heterogeneity evaluation. To investigate the value of advanced multiparametric MR imaging biomarker analysis based on radiomic features and machine learning classification, in the non-invasive evaluation of tumor heterogeneity towards the differentiation of Low Grade vs. High Grade Gliomas. Forty histologically confirmed glioma patients (20 LGG and 20 HGG) who underwent a standard 3T-MRI tumor protocol with conventional (T1 pre/post-contrast, T2-FSE, T2-FLAIR) and advanced techniques (Diffusion Tensor and Perfusion Imaging, 1H-MR Spectroscopy), were included. A semi-automated segmentation technique, based on T1W-C and DTI, was used for tumor core delineation in all available parametric maps. 3D Texture analysis considered 12 Histogram, 11 Co-Occurrence Matrix (GLCM) and 5 Run Length Matrix (GLRLM) features, derived from p, q, MD, FA, T1W-C, T2W-FSE, T2W-FLAIR and raw DSCE data. Along with 1H-MRS metabolic ratios and mean rCBV values, a total of 581 attributes for each subject were obtained. A Support Vector Machine – Recursive Feature Elimination (SVM-RFE) algorithm and SVM classifier were utilized for feature selection and classification, respectively. Three different SVM classifiers were evaluated with consecutively SVM-RFE feature subsets. Linear SMO classifier demonstrated the highest performance for determining the optimal feature subset. Finally, 21 SVM-RFE top-ranked features were adopted, for training and testing the SMO classifier with leave-one-out cross-validation, achieving 95.5% Accuracy, 95% Sensitivity, 96% Specificity and 95.5% Area Under ROC Curve. Results demonstrate that quantitative analysis of phenotypic characteristics, based on advanced multiparametric MR neuroimaging data and texture features, utilizing state-of-the-art radiomic analysis methods, can significantly contribute to the pre-treatment glioma grade differentiation. [ABSTRACT FROM AUTHOR]

Background: Accurate glioma grading plays an important role in the clinical management of patients and is also the basis of molecular stratification nowadays.Purpose/hypothesis: To verify the superiority of radiomics features extracted from multiparametric MRI to glioma grading and evaluate the grading potential of different MRI sequences or parametric maps.Study Type: Retrospective; radiomics.Population: A total of 153 patients including 42, 33, and 78 patients with Grades II, III, and IV gliomas, respectively.Field Strength/sequence: 3.0T MRI/T1 -weighted images before and after contrast-enhanced, T2 -weighted, multi-b-value diffusion-weighted and 3D arterial spin labeling images.Assessment: After multiparametric MRI preprocessing, high-throughput features were derived from patients' volumes of interests (VOIs). The support vector machine-based recursive feature elimination was adopted to find the optimal features for low-grade glioma (LGG) vs. high-grade glioma (HGG), and Grade III vs. IV glioma classification tasks. Then support vector machine (SVM) classifiers were established using the optimal features. The accuracy and area under the curve (AUC) was used to assess the grading efficiency.Statistical Tests: Student's t-test or a chi-square test were applied on different clinical characteristics to confirm whether intergroup significant differences exist.Results: Patients' ages between LGG and HGG groups were significantly different (P < 0.01). For each patient, 420 texture and 90 histogram parameters were derived from 10 VOIs of multiparametric MRI. SVM models were established using 30 and 28 optimal features for classifying LGGs from HGGs and grades III from IV, respectively. The accuracies/AUCs were 96.8%/0.987 for classifying LGGs from HGGs, and 98.1%/0.992 for classifying grades III from IV, which were more promising than using histogram parameters or using the single sequence MRI.Data Conclusion: Texture features were more effective for noninvasively grading gliomas than histogram parameters. The combined application of multiparametric MRI provided a higher grading efficiency. The proposed radiomic strategy could facilitate clinical decision-making for patients with varied glioma grades.Level Of Evidence: 3 Technical Efficacy: Stage 2 J. Magn. Reson. Imaging 2018;48:1518-1528. [ABSTRACT FROM AUTHOR]

Background and purpose: Gliomas are very heterogeneous tumors, glioma grading is currently based on the histologic assessment. With noninvasive measurement of vascular permeability by CT Perfusion (CTP) multiple perfusion parameters which can be obtained with a single acquisition the aim of this study was to find the most sensitive and specific CTP parameters and their cutoffs that can be used to differentiate between low and high grade brain gliomas. Material and methods: Twenty-five patients were included in this study divided into two groups: group A includes 15 patients with high grade glioma, group B includes 10 patients with low grade glioma; CTP was done for all patients, perfusion values of tumors were then calculated, and statistical analysis was done using IBM SPSS. Results: A statistically highly significant difference was found between the two groups regarding the BF, BV and PS with P values of 0.0003, 0.00026 and 0.0009 respectively. The two groups were found to be self-discriminated by BV and PS with sensitivity of 95% and 86% respectively. Conclusion: CTP shows high sensitivity in terms of differentiation between high and low grade adult gliomas.

Background: Accurate glioma grading before surgery is of the utmost importance in treatment planning and prognosis prediction. But previous studies on magnetic resonance imaging (MRI) images were not effective enough. According to the remarkable performance of convolutional neural network (CNN) in medical domain, we hypothesized that a deep learning algorithm can achieve high accuracy in distinguishing the World Health Organization (WHO) low grade and high grade gliomas.Methods: One hundred and thirteen glioma patients were retrospectively included. Tumor images were segmented with a rectangular region of interest (ROI), which contained about 80% of the tumor. Then, 20% data were randomly selected and leaved out at patient-level as test dataset. AlexNet and GoogLeNet were both trained from scratch and fine-tuned from models that pre-trained on the large scale natural image database, ImageNet, to magnetic resonance images. The classification task was evaluated with five-fold cross-validation (CV) on patient-level split.Results: The performance measures, including validation accuracy, test accuracy and test area under curve (AUC), averaged from five-fold CV of GoogLeNet which trained from scratch were 0.867, 0.909, and 0.939, respectively. With transfer learning and fine-tuning, better performances were obtained for both AlexNet and GoogLeNet, especially for AlexNet. Meanwhile, GoogLeNet performed better than AlexNet no matter trained from scratch or learned from pre-trained model.Conclusion: In conclusion, we demonstrated that the application of CNN, especially trained with transfer learning and fine-tuning, to preoperative glioma grading improves the performance, compared with either the performance of traditional machine learning method based on hand-crafted features, or even the CNNs trained from scratch.

Background Grading of gliomas is critical information related to prognosis and survival. We aimed to apply a radiomics approach using various machine learning classifiers to determine the glioma grading. Methods We considered 285 (high grade n = 210, low grade n = 75) cases obtained from the Brain Tumor Segmentation 2017 Challenge. Manual annotations of enhancing tumors, non-enhancing tumors, necrosis, and edema were provided by the database. Each case was multi-modal with T1-weighted, T1-contrast enhanced, T2-weighted, and FLAIR images. A five-fold cross validation was adopted to separate the training and test data. A total of 468 radiomics features were calculated for three types of regions of interest. The minimum redundancy maximum relevance algorithm was used to select features useful for classifying glioma grades in the training cohort. The selected features were used to build three classifier models of logistics, support vector machines, and random forest classifiers. The classification performance of the models was measured in the training cohort using accuracy, sensitivity, specificity, and area under the curve (AUC) of the receiver operating characteristic curve. The trained classifier models were applied to the test cohort. Results Five significant features were selected for the machine learning classifiers and the three classifiers showed an average AUC of 0.9400 for training cohorts and 0.9030 (logistic regression 0.9010, support vector machine 0.8866, and random forest 0.9213) for test cohorts. Discussion Glioma grading could be accurately determined using machine learning and feature selection techniques in conjunction with a radiomics approach. The results of our study might contribute to high-throughput computer aided diagnosis system for gliomas.

Purpose: To evaluate the feasibility of a simplified method based on diffusion-weighted imaging (DWI) acquired with three b-values to measure tissue perfusion linked to microcirculation, to validate it against from perfusion-related parameters derived from intravoxel incoherent motion (IVIM) and dynamic contrast-enhanced (DCE) magnetic resonance (MR) imaging, and to investigate its utility to differentiate low- from high-grade gliomas.Materials and Methods: The prospective study was approved by the local institutional review board and written informed consent was obtained from all patients. From May 2016 and May 2017, 50 patients confirmed with glioma were assessed with multi-b-value DWI and DCE MR imaging at 3.0 T. Besides conventional apparent diffusion coefficient (ADC0,1000) map, perfusion-related parametric maps for IVIM-derived perfusion fraction (f) and pseudodiffusion coefficient (D*), DCE MR imaging-derived pharmacokinetic metrics, including Ktrans, ve and vp, as well as a metric named simplified perfusion fraction (SPF), were generated. Correlation between perfusion-related parameters was analyzed by using the Spearman rank correlation. All imaging parameters were compared between the low-grade (n = 19) and high-grade (n = 31) groups by using the Mann-Whitney U test. The diagnostic performance for tumor grading was evaluated with receiver operating characteristic (ROC) analysis.Results: SPF showed strong correlation with IVIM-derived f and D* (ρ = 0.732 and 0.716, respectively; both P < 0.001). Compared with f, SPF was more correlated with DCE MR imaging-derived Ktrans (ρ = 0.607; P < 0.001) and vp (ρ = 0.397; P = 0.004). Among all parameters, SPF achieved the highest accuracy for differentiating low- from high-grade gliomas, with an area under the ROC curve value of 0.942, which was significantly higher than that of ADC0,1000 (P = 0.004). By using SPF as a discriminative index, the diagnostic sensitivity and specificity were 87.1% and 94.7%, respectively, at the optimal cut-off value of 19.26%.Conclusion: The simplified method to measure tissue perfusion based on DWI by using three b-values may be helpful to differentiate low- from high-grade gliomas. SPF may serve as a valuable alternative to measure tumor perfusion in gliomas in a noninvasive, convenient and efficient way.

Accurate grading of brain tumors plays a crucial role in the diagnosis and treatment of glioma. While convolutional neural networks (CNNs) have shown promising performance in this task, their clinical applicability is still constrained by the interpretability and robustness of the models. In the conventional framework, the classification model is trained first, and then visual explanations are generated. However, this approach often leads to models that prioritize classification performance or complexity, making it difficult to achieve a precise visual explanation. Motivated by these challenges, we propose the Unified Visualization and Classification Network (UniVisNet), a novel framework that aims to improve both the classification performance and the generation of high-resolution visual explanations. UniVisNet addresses attention misalignment by introducing a subregion-based attention mechanism, which replaces traditional down-sampling operations. Additionally, multiscale feature maps are fused to achieve higher resolution, enabling the generation of detailed visual explanations. To streamline the process, we introduce the Unified Visualization and Classification head (UniVisHead), which directly generates visual explanations without the need for additional separation steps. Through extensive experiments, our proposed UniVisNet consistently outperforms strong baseline classification models and prevalent visualization methods. Notably, UniVisNet achieves remarkable results on the glioma grading task, including an AUC of 94.7%, an accuracy of 89.3%, a sensitivity of 90.4%, and a specificity of 85.3%. Moreover, UniVisNet provides visually interpretable explanations that surpass existing approaches. In conclusion, UniVisNet innovatively generates visual explanations in brain tumor grading by simultaneously improving the classification performance and generating high-resolution visual explanations. This work contributes to the clinical application of deep learning, empowering clinicians with comprehensive insights into the spatial heterogeneity of glioma., Competing Interests: Declaration of competing interest None Declared, (Copyright © 2023 Elsevier Ltd. All rights reserved.)

The aim of the present study was to assess the value of susceptibility-weighted imaging (SWI) and diffusion-weighted imaging (DWI) in the grading of gliomas and to evaluate the correlation between these quantitative parameters derived from SWI and DWI. A total of 49 patients with glioma were assessed by DWI and SWI. The evaluation included the ratio of apparent diffuse coefficient values between the solid portion of tumors and contralateral normal white matter (rADC) and the degree of intratumoral susceptibility signal intensity (ITSS) within tumors. Receiver operating characteristic curve (ROC) analyses were performed and the area under the ROC curve was calculated to compare the diagnostic performance, determine optimum thresholds for tumor grading, and calculate the sensitivity, specificity, positive predictive value (PPV) and negative predictive value (NPV) for identifying high-grade gliomas. The correlation between DWI- and SWI-derived parameters was also evaluated. The rADC and the degrees of ITSS within tumors were significantly higher in high-grade gliomas than those in low-grade gliomas. ROC curve analysis indicated that the rADC was a better index for grading gliomas than the ITSS degree. Statistical analysis demonstrated a threshold value of 1.497 for rADC to provide a sensitivity, specificity, PPV and NPV of 86.2, 85.0, 89.3 and 81.0%, respectively, for determining high-grade gliomas. A degree of ITSS of 1.5 was defined as the threshold to identify high-grade gliomas and sensitivity, specificity, PPV and NPV of 82.8, 75.0, 82.8 and 75.0% were obtained, respectively. Furthermore, a moderate inverse correlation between rADC and the ITSS degree was revealed. Combination of SWI with DWI may provide valuable information for glioma grading. [ABSTRACT FROM AUTHOR]

Purpose: To evaluate the feasibility of a simplified method based on diffusion-weighted imaging (DWI) acquired with three b-values to measure tissue perfusion linked to microcirculation, to validate it against from perfusion-related parameters derived from intravoxel incoherent motion (IVIM) and dynamic contrast-enhanced (DCE) magnetic resonance (MR) imaging, and to investigate its utility to differentiate low-from high-grade gliomas. Materials and Methods: The prospective study was approved by the local institutional review board and written informed consent was obtained from all patients. From May 2016 and May 2017, 50 patients confirmed with glioma were assessed with multib-value DWI and DCE MR imaging at 3.0 T. Besides conventional apparent diffusion coefficient (ADC0,1000) map, perfusion-related parametric maps for IVIM-derived perfusion fraction (f) and pseudodiffusion coefficient (D+), DCE MR imaging-derived pharmacokinetic metrics, including Ktrans, ve and vp, as well as a metric named simplified perfusion fraction (SPF), were generated. Correlation between perfusion-related parameters was analyzed by using the Spearman rank correlation. All imaging parameters were compared between the low-grade (n = 19) and high-grade (n = 31) groups by using the Mann-Whitney U test. The diagnostic performance for tumor grading was evaluated with receiver operating characteristic (ROC) analysis. Results: SPF showed strong correlation with IVIM-derived f and D+ (p = 0.732 and 0.716, respectively; both P < 0.001). Compared with f, SPF was more correlated with DCE MR imaging-derived Ktrans (p = 0.607; P < 0.001) and vp (p = 0.397; P = 0.004). Among all parameters, SPF achieved the highest accuracy for differentiating low-from high-grade gliomas, with an area under the ROC curve value of 0.942, which was significantly higher than that of ADC0,1000 (P = 0.004). By using SPF as a discriminative index, the diagnostic sensitivity and specificity were 87.1% and 94.7%, respectively, at the optimal cut-off value of 19.26%. Conclusion: The simplified method to measure tissue perfusion based on DWI by using three b-values may be helpful to differentiate low-from high-grade gliomas. SPF mayserve as a valuable alternative to measure tumor perfusion in gliomas in a noninvasive, convenient and efficient way. [ABSTRACT FROM AUTHOR]

Background: Dynamic-contrast-enhanced (DCE) MRI data acquired using gradient echo based sequences is affected by errors in flip angle (FA) due to transmit B1 inhomogeneity (B1inh). The purpose of the study was to evaluate the effect of B1inh on quantitative analysis of DCE-MRI data of human brain tumor patients and to evaluate the clinical significance of B1inh correction of perfusion parameters (PPs) on tumor grading.Methods: An MRI study was conducted on 35 glioma patients at 3T. The patients had histologically confirmed glioma with 23 high-grade (HG) and 12 low-grade (LG). Data for B1-mapping, T1-mapping and DCE-MRI were acquired. Relative B1 maps (B1rel) were generated using the saturated-double-angle method. T1-maps were computed using the variable flip-angle method. Post-processing was performed for conversion of signal-intensity time (S(t)) curve to concentration-time (C(t)) curve followed by tracer kinetic analysis (Ktrans, Ve, Vp, Kep) and first pass analysis (CBV, CBF) using the general tracer-kinetic model. DCE-MRI data was analyzed without and with B1inh correction and errors in PPs were computed. Receiver-operating-characteristic (ROC) analysis was performed on HG and LG patients. Simulations were carried out to understand the effect of B1 inhomogeneity on DCE-MRI data analysis in a systematic way. S(t) curves mimicking those in tumor tissue, were generated and FA errors were introduced followed by error analysis of PPs. Dependence of FA-based errors on the concentration of contrast agent and on the duration of DCE-MRI data was also studied. Simulations were also done to obtain Ktrans of glioma patients at different B1rel values and see whether grading is affected or not.Results: Current study shows that B1rel value higher than nominal results in an overestimation of C(t) curves as well as derived PPs and vice versa. Moreover, at same B1rel values, errors were large for larger values of C(t). Simulation results showed that grade of patients can change because of B1inh.Conclusions: B1inh in the human brain at 3T-MRI can introduce substantial errors in PPs derived from DCE-MRI data that might affect the accuracy of tumor grading, particularly for border zone cases. These errors can be mitigated using B1inh correction during DCE-MRI data analysis. [ABSTRACT FROM AUTHOR]

Purpose Aim of this retrospective study was to compare diagnostic accuracy of proposed automatic normalization method to quantify the relative cerebral blood volume (rCBV) with existing contra-lateral region of interest (ROI) based CBV normalization method for glioma grading using T1-weighted dynamic contrast enhanced MRI (DCE-MRI). Material and methods Sixty patients with histologically confirmed gliomas were included in this study retrospectively. CBV maps were generated using T1-weighted DCE-MRI and are normalized by contralateral ROI based method (rCBV_contra), unaffected white matter (rCBV_WM) and unaffected gray matter (rCBV_GM), the latter two of these were generated automatically. An expert radiologist with > 10 years of experience in DCE-MRI and a non-expert with one year experience were used independently to measure rCBVs. Cutoff values for glioma grading were decided from ROC analysis. Agreement of histology with rCBV_WM, rCBV_GM and rCBV_contra respectively was studied using Kappa statistics and intra-class correlation coefficient (ICC). Result The diagnostic accuracy of glioma grading using the measured rCBV_contra by expert radiologist was found to be high (sensitivity = 1.00, specificity = 0.96, p < 0.001) compared to the non-expert user (sensitivity = 0.65, specificity = 0.78, p < 0.001). On the other hand, both the expert and non-expert user showed similar diagnostic accuracy for automatic rCBV_WM (sensitivity = 0.89, specificity = 0.87, p = 0.001) and rCBV_GM (sensitivity = 0.81, specificity = 0.78, p = 0.001) measures. Further, it was also observed that, contralateral based method by expert user showed highest agreement with histological grading of tumor (kappa = 0.96, agreement 98.33%, p < 0.001), however; automatic normalization method showed same percentage of agreement for both expert and non-expert user. rCBV_WM showed an agreement of 88.33% (kappa = 0.76,p < 0.001) with histopathological grading. Conclusion It was inferred from this study that, in the absence of expert user, automated normalization of CBV using the proposed method could provide better diagnostic accuracy compared to the manual contralateral based approach. [ABSTRACT FROM AUTHOR]

The fusion of multi-modal data, e.g., medical images and genomic profiles, can provide complementary information and further benefit disease diagnosis. However, multi-modal disease diagnosis confronts two challenges: (1) how to produce discriminative multi-modal representations by exploiting complementary information while avoiding noisy features from different modalities. (2) how to obtain an accurate diagnosis when only a single modality is available in real clinical scenarios. To tackle these two issues, we present a two-stage disease diagnostic framework. In the first multi-modal learning stage, we propose a novel Momentum-enriched Multi-Modal Low-Rank (M 3 LR) constraint to explore the high-order correlations and complementary information among different modalities, thus yielding more accurate multi-modal diagnosis. In the second stage, the privileged knowledge of the multi-modal teacher is transferred to the unimodal student via our proposed Discrepancy Supervised Contrastive Distillation (DSCD) and Gradient-guided Knowledge Modulation (GKM) modules, which benefit the unimodal-based diagnosis. We have validated our approach on two tasks: (i) glioma grading based on pathology slides and genomic data, and (ii) skin lesion classification based on dermoscopy and clinical images. Experimental results on both tasks demonstrate that our proposed method consistently outperforms existing approaches in both multi-modal and unimodal diagnoses. • We present a two-stage disease diagnostic framework that first trains a multi-modal network and then distills the multi-modal knowledge to train a unimodal network. • In the multi-modal training stage, we devise a novel Momentum-enriched Multi-Modal Low-Rank ( M 3 L R) constraint to capture the consensus and high-order complementary information among different modalities and momentum encoders. • In the unimodal training stage, we propose a novel Discrepancy Supervised Contrastive Distillation (DSCD) paradigm and a Gradient-guided Knowledge Modulation (GKM) scheme to transfer the multi-modal knowledge toward a more accurate unimodal diagnosis. • The DSCD module distills rich structural knowledge by pulling multiple positive pairs (from the same class) with teacher–student discrepancy while pushing apart negative pairs from different classes. The GKM constructs a comprehensive knowledge bank and adaptively modulates the contributions of multiple knowledge according to their agreement in the gradient space. • We perform extensive experiments on the glioma grading and skin lesion classification tasks to validate the effectiveness of the proposed framework and the experimental results demonstrate the superiority of the proposed method over existing approaches in both multi-modal and unimodal diagnoses. [ABSTRACT FROM AUTHOR]

• An expert knowledge guided manifold representation learning framework is proposed for accurate glioma grading. • The framework can make direct grading without human annotations during testing. • The framework is designed in a physician-friendly way and is flexible to be integrated with different network architectures. Radiomics and deep learning have shown high popularity in automatic glioma grading. Radiomics extracts handcrafted features that quantitatively describe the expert knowledge of gliomas, and deep learning is powerful in extracting numerous learned features that facilitate the final classification. However, the performance of existing methods can still be improved as their complementary strengths haven't been sufficiently investigated and integrated. Furthermore, the final prediction at the testing phase normally needs lesion maps which rely too much on the segmentation accuracy and could be troublesome. To address these challenges, we propose an expert knowledge guided manifold representation learning (ENROL) framework for accurate glioma grading without the reliance on segmentation maps during testing. Low-dimensional manifolds of the handcrafted features and learned features are constructed to mine the implicit relationship between radiomics and deep learning, and therefore to dig mutual consent representation for the glioma grades. With a specially designed manifold discrepancy measurement, the grading model can exploit the MRI data and expert knowledge more effectively and get rid of the requirement of lesion segmentation maps at the testing phase. The proposed framework is flexible regarding deep learning architectures to be utilized. Three different architectures have been evaluated on both single-sequence and multi-sequence MRI data. Results show that our framework can generate promising glioma grading performance. [ABSTRACT FROM AUTHOR]

Introduction: Preoperative tumour grading is imperative owing to difference in invasive, aggressive tendencies of different grades of glial tumours implying varied prognosis, therapeutic options. Histopathological examination has inherent sampling errors. Magnetic Resonance Spectroscopy (MRS) and Diffusion Weighted Imaging (DWI) can provide non invasive information about internal mileu hence, aiding in tumour grading by adding to information provided by conventional MRI sequences. Aim: To evaluate the role of multivoxel intermediate TE 2D CSI MRS and 2D echoplanar diffusion imaging in grading of primary glial brain tumours. Material and Methods: A prospective study was conducted in Department of Radiology, Teerthanker Mahaveer Medical College and Research Centre, Uttar Pradesh, India, from April 2015 to August 2016 after obtaining necessary approvals from Institutional Ethical Committee and written informed consent from all participants on histopathological proven cases of glial brain tumours that underwent multivoxel MRS using intermediate TE 2D chemical shift imaging and DWI using 2D echoplanar imaging. Tumour grade calculated on MRI using MRS and DWI was compared with histopathological grading. Positive Predictive Value (PPV), Negative Predictive Value (NPV), Sensitivity, specificity and accuracy were calculated for each parameter and statistical significance was evaluated using two tailed Pearson test. Results: Choline: N Acetyl aspartate (Cho: NAA) and Choline: creatinine (Cho: Cr) ratios from MRS as well as Apparent Diffusion Coffecient (ADC) values from DWI were significantly higher with increasing severity of tumour grade. Accuracy of 58.6% was obtained with DWI while it was 83% with MRS. MRS and DWI used together provided 88.4% accuracy. All parameters evaluated showed statistical significance. Conclusion: Both DWI as well as MRS were found to have statistically significant roles in grading of glial brain tumours. MRS was found to be more useful than DWI.

Gliomas are the most common intra-axial primary brain tumour; therefore, predicting glioma grade would influence therapeutic strategies. Although several methods based on single or multiple parameters from diagnostic images exist, a definitive method for pre-operatively determining glioma grade remains unknown. We aimed to develop an unsupervised method using multiple parameters from pre-operative diffusion tensor images for obtaining a clustered image that could enable visual grading of gliomas. Fourteen patients with low-grade gliomas and 19 with high-grade gliomas underwent diffusion tensor imaging and three-dimensional T1-weighted magnetic resonance imaging before tumour resection. Seven features including diffusion-weighted imaging, fractional anisotropy, first eigenvalue, second eigenvalue, third eigenvalue, mean diffusivity and raw T2 signal with no diffusion weighting, were extracted as multiple parameters from diffusion tensor imaging. We developed a two-level clustering approach for a self-organizing map followed by the K-means algorithm to enable unsupervised clustering of a large number of input vectors with the seven features for the whole brain. The vectors were grouped by the self-organizing map as protoclusters, which were classified into the smaller number of clusters by K-means to make a voxel-based diffusion tensor-based clustered image. Furthermore, we also determined if the diffusion tensor-based clustered image was really helpful for predicting pre-operative glioma grade in a supervised manner. The ratio of each class in the diffusion tensor-based clustered images was calculated from the regions of interest manually traced on the diffusion tensor imaging space, and the common logarithmic ratio scales were calculated. We then applied support vector machine as a classifier for distinguishing between low- and high-grade gliomas. Consequently, the sensitivity, specificity, accuracy and area under the curve of receiver operating characteristic curves from the 16-class diffusion tensor-based clustered images that showed the best performance for differentiating high- and low-grade gliomas were 0.848, 0.745, 0.804 and 0.912, respectively. Furthermore, the log-ratio value of each class of the 16-class diffusion tensor-based clustered images was compared between low- and high-grade gliomas, and the log-ratio values of classes 14, 15 and 16 in the high-grade gliomas were significantly higher than those in the low-grade gliomas (p

Background: Distinguishing low-grade from high-grade gliomas can aid in optimal treatment planning and prognostication. Diffusion-weighted imaging (DWI) and magnetic resonance spectroscopy (MRS) have been applied in several studies for noninvasive glioma grading. However, these studies focused on limited aspects of these imaging techniques and used different study setups, resulting in occasionally inconsistent and incomparable conclusions in the literature. Objectives: This study was designed to introduce the optimal imaging setup and the most reliable and applicable imaging parameters in glioma grading, using DWI and MRS. Patients and Methods: During this prospective study, using a 3T-MR scanner, 55 glioma patients underwent brain MRS with short, intermediate, and long echo times (TEs), as well as DWI using low, intermediate, and high b-values. Postoperatively, all of the specimens were graded pathologically using light microscopy. Results:We found that Max (Chol/Cr)/ Min (NAA/Cr) ((maximum choline to creatine ratio)/(minimum N-acetyl aspartate to creatine ratio)), followed by Max (Chol/ Cr), both in long-TE, were the most reliable metabolite ratios on MRS for accurate glioma grading. These had values for area under the curve (AUC) of 0.92 (P < 0.05) and 0.89 (P = 0.001), respectively, compared to conventional MR imaging (cMRI), which had an AUC of 0.83 (P < 0.05). DWI at maximal accuracy showed an AUC of 0.80 (P < 0.05). Conclusion: Max (Chol/Cr)/Min (NAA/Cr) in long-TE was the most reliable of all of the MRS parameters studied, while DWI showed no superiority over cMRI in glioma grading. No significant differences existed among the various b-values applied, or between the minimum and mean tumor apparent diffusion coefficient values used in DWI-based glioma grading. [ABSTRACT FROM AUTHOR]

To noninvasively assess the diagnostic performance of diffusion-weighted imaging (DWI), bi-exponential intravoxel incoherent motion imaging (IVIM) and three-dimensional pseudo-continuous arterial spin labeling (3D pCASL) in differentiating lower-grade gliomas (LGGs) from high-grade gliomas (HGGs), and predicting the isocitrate dehydrogenase (IDH) mutation status. Ninety-five patients with pathologically confirmed grade 2–4 gliomas with preoperative DWI, IVIM and 3D pCASL were enrolled in this study. The Student's t test and Mann-Whitney U test were used to evaluate differences in parameters of DWI, IVIM and 3D pCASL between LGG and HGG as well as between mutant and wild-type IDH in grade 2 and 3 diffusion astrocytoma; receiver operator characteristic (ROC) analysis was used to assess the diagnostic performance. The value of ADC mean , ADC min , D mean and D min in HGGs were lower than in LGGs, while the value of CBF mean and CBF max in HGGs were higher than in LGGs. In ROC analysis, the AUC values of D mean , D min and CBF max were 0.827, 0.878 and 0.839, respectively. The combination of CBF max and D min displayed the highest diagnostic performance to distinguish LGGs from HGGs, with AUC 0.906, sensitivity 82.4 %, and specificity 86.4 %. In grades 2 and 3 diffusion astrocytoma patients, ADC min , D mean , D min , CBF mean and CBF max showed significant differences between IDH mut and IDH wt group (p < 0.05, 0.001, 0.001, 0.01 and 0.001, respectively) and the AUC values were 0. 709, 0.849, 0.919, 0.755 and 0.873, respectively. Similarly, the combination of CBF max and D min demonstrated the highest AUC value (0.938) in prediction IDH mutation status, with sensitivity 92.9 %, and specificity 95.5 %. The combination of IVIM and 3D pCASL can be used in prediction histologic grade and IDH mutation status of glioma noninvasively. [ABSTRACT FROM AUTHOR]

• A reliable feature extraction pipeline for multiparametric MRI data was described. • Image preprocessing is a fundamental step in radiomic MRI analysis. • A set of robust features respect to the image preprocessing technique and features extraction settings was founded. • The robustness of radiomic features and the impact on their predictive power were investigated. Analysis pipelines based on the computation of radiomic features on medical images are widely used exploration tools across a large variety of image modalities. This study aims to define a robust processing pipeline based on Radiomics and Machine Learning (ML) to analyze multiparametric Magnetic Resonance Imaging (MRI) data to discriminate between high-grade (HGG) and low-grade (LGG) gliomas. The dataset consists of 158 multiparametric MRI of patients with brain tumor publicly available on The Cancer Imaging Archive, preprocessed by the BraTS organization committee. Three different types of image intensity normalization algorithms were applied and 107 features were extracted for each tumor region, setting the intensity values according to different discretization levels. The predictive power of radiomic features in the LGG versus HGG categorization was evaluated by using random forest classifiers. The impact of the normalization techniques and of the different settings in the image discretization was studied in terms of the classification performances. A set of MRI-reliable features was defined selecting the features extracted according to the most appropriate normalization and discretization settings. The results show that using MRI-reliable features improves the performance in glioma grade classification (AUC = 0.93 ± 0.05) with respect to the use of raw (AUC = 0.88 ± 0.08) and robust features (AUC = 0.83 ± 0.08), defined as those not depending on image normalization and intensity discretization. These results confirm that image normalization and intensity discretization strongly impact the performance of ML classifiers based on radiomic features. Thus, special attention should be provided in the image preprocessing step before typical radiomic and ML analysis are carried out. [ABSTRACT FROM AUTHOR]

Introduction: In this work, we aim to assess the significance of diffusion tensor imaging (DTI) and diffusion kurtosis imaging (DKI) parameters in grading gliomas. Methods: Retrospective studies were performed on 53 subjects with gliomas belonging to WHO grade II ( n = 19), grade III ( n = 20) and grade IV ( n = 14). Expert marked regions of interest (ROIs) covering the tumour on T2-weighted images. Statistical texture measures such as entropy and busyness calculated over ROIs on diffusion parametric maps were used to assess the tumour heterogeneity. Additionally, we propose a volume heterogeneity index derived from cross correlation (CC) analysis as a tool for grading gliomas. The texture measures were compared between grades by performing the Mann-Whitney test followed by receiver operating characteristic (ROC) analysis for evaluating diagnostic accuracy. Results: Entropy, busyness and volume heterogeneity index for all diffusion parameters except fractional anisotropy and anisotropy of kurtosis showed significant differences between grades. The Mann-Whitney test on mean diffusivity (MD), among DTI parameters, resulted in the highest discriminability with values of P = 0.029 (0.0421) for grade II vs. III and P = 0.0312 (0.0415) for III vs. IV for entropy (busyness). In DKI, mean kurtosis (MK) showed the highest discriminability, P = 0.018 (0.038) for grade II vs. III and P = 0.022 (0.04) for III vs. IV for entropy (busyness). Results of CC analysis illustrate the existence of homogeneity in volume (uniformity across slices) for lower grades, as compared to higher grades. Hypothesis testing performed on volume heterogeneity index showed P values of 0.0002 (0.0001) and 0.0003 (0.0003) between grades II vs. III and III vs. IV, respectively, for MD (MK). Conclusion: In summary, the studies demonstrated great potential towards automating grading gliomas by employing tumour heterogeneity measures on DTI and DKI parameters. [ABSTRACT FROM AUTHOR]

Objectives To assess the diagnostic accuracy of DW-MRI, ADC value, and MRS in preoperative glioma grading because of its importance in treatment planning. Patients and methods A prospective study included 30 patients with gliomas, based on CT and cMRI findings, referred from Neurosurgery Department for DWI and MRS. Results were correlated with histopathological diagnosis after surgical resection. Results ADC values were significantly higher in low grade gliomas relative to high grade ones. The lowest value was for GBM, 100% sensitivity, specificity, PPV and NPP in glioma grading. MRS revealed higher Cho/NAA and Cho/Cr ratios in high grade neoplasms and characteristic elevated lipid peak with statistical significant difference between low grade and high grade gliomas. MRS was more accurate than ADC value in detecting peritumoral infiltration of high grade gliomas, but there were no statistically significant differences between anaplastic and GBM in tumoral and peritumoral regions. Conclusion DW-MRI had higher sensitivity, specificity and accuracy than cMRI and MRS in glioma grading while MRS is more accurate than ADC value in assessing peri-tumoral infiltration based on high metabolite ratios in peri-tumoral tissue for anaplastic glioma and GBM, but there were no statistically significant differences between high grade groups. [ABSTRACT FROM AUTHOR]

Purpose. To determine glioma grading by applying radiomic analysis or deep convolutional neural networks (DCNN) and to benchmark both approaches on broader validation sets. Methods. Seven public datasets were considered: (1) low-grade glioma or high-grade glioma (369 patients, BraTS'20) (2) well-differentiated liposarcoma or lipoma (115, LIPO); (3) desmoid-type fibromatosis or extremity soft-tissue sarcomas (203, Desmoid); (4) primary solid liver tumors, either malignant or benign (186, LIVER); (5) gastrointestinal stromal tumors (GISTs) or intra-abdominal gastrointestinal tumors radiologically resembling GISTs (246, GIST); (6) colorectal liver metastases (77, CRLM); and (7) lung metastases of metastatic melanoma (103, Melanoma). Radiomic analysis was performed on 464 (2016) radiomic features for the BraTS'20 (others) datasets respectively. Random forests (RF), Extreme Gradient Boosting (XGBOOST) and a voting algorithm comprising both classifiers were tested. The parameters of the classifiers were optimized using a repeated nested stratified cross-validation process. The feature importance of each classifier was computed using the Gini index or permutation feature importance. DCNN was performed on 2D axial and sagittal slices encompassing the tumor. A balanced database was created, when necessary, using smart slices selection. ResNet50, Xception, EficientNetB0, and EfficientNetB3 were transferred from the ImageNet application to the tumor classification and were fine-tuned. Five-fold stratified cross-validation was performed to evaluate the models. The classification performance of the models was measured using multiple indices including area under the receiver operating characteristic curve (AUC). Results. The best radiomic approach was based on XGBOOST for all datasets; AUC was 0.934 (BraTS'20), 0.86 (LIPO), 0.73 (LIVER), (0.844) Desmoid, 0.76 (GIST), 0.664 (CRLM), and 0.577 (Melanoma) respectively. The best DCNN was based on EfficientNetB0; AUC was 0.99 (BraTS'20), 0.982 (LIPO), 0.977 (LIVER), (0.961) Desmoid, 0.926 (GIST), 0.901 (CRLM), and 0.89 (Melanoma) respectively. Conclusion. Tumor classification can be accurately determined by adapting state-of-the-art machine learning algorithms to the medical context., (© 2023 IOP Publishing Ltd.)

Glioma grading is critical for treatment selection, and the fine classification between glioma grades II and III is still a pathological challenge. Traditional systems based on a single deep learning (DL) model can only show relatively low accuracy in distinguishing glioma grades II and III. Introducing ensemble DL models by combining DL and ensemble learning techniques, we achieved annotation-free glioma grading (grade II or III) from pathological images. We established multiple tile-level DL models using residual network ResNet-18 architecture and then used DL models as component classifiers to develop ensemble DL models to achieve patient-level glioma grading. Whole-slide images of 507 subjects with low-grade glioma (LGG) from the Cancer Genome Atlas (TCGA) were included. The 30 DL models exhibited an average area under the curve (AUC) of 0.7991 in patient-level glioma grading. Single DL models showed large variation, and the median between-model cosine similarity was 0.9524, significantly smaller than the threshold of 1.0. The ensemble model based on logistic regression (LR) methods with a 14-component DL classifier (LR-14) demonstrated a mean patient-level accuracy and AUC of 0.8011 and 0.8945, respectively. Our proposed LR-14 ensemble DL model achieved state-of-the-art performance in glioma grade II and III classifications based on unannotated pathological images., (© 2023 The Authors.)

Purpose Dynamic contrast enhanced (DCE) MRI is used to grade and to monitor the progression of glioma while on treatment. Usually, a fixed hematocrit (Hct) value for adults is assumed to be ~ 45%; however, it is actually known for individual variations. Purpose of this study was to investigate the effect of measured Hct values in glioma grading using DCE-MRI. Materials and methods Fifty glioma patients were included in this study. Kinetic and hemodynamic parameters were estimated for each patient using assumed as well as measured Hct values. To look the changes in Hct value over time, Hct was measured multiple times from 10 of these glioma patients who were on treatment. Simulation was done to look for the effect of extreme variations of Hct values on perfusion metrics. The data was compared to look for significant differences in the perfusion metrics derived from assumed and measured Hct values. Results The measured Hct value in patients was found to be (40.4 ± 4.28)%. The sensitivity and specificity of DCE-MRI parameters in glioma grading were not significantly influenced by using measured vis-a-vis assumed Hct values. The serial Hct values from 10 patients who were on treatment showed a fluctuation of 15–20% over time. The simulated data showed linear influence of Hct values on kinetic parameters. The tumor grading was altered on altering the Hct values in borderline cases. Conclusion Hct values influence the hemodynamic and kinetic metrics linearly and may affect glioma grading. However, perfusion metrics values might change significantly with large change in Hct values, especially in patients who are on chemotherapy necessitating its use in the DCE model. [ABSTRACT FROM AUTHOR]

Purpose: Current guidelines for glioma imaging by positron emission tomography (PET) using the amino acid analogue O-(2-[F]fluoroethyl)-L-tyrosine (F-FET) recommend image acquisition from 20-40 min post injection (p.i.). The maximal tumour-to-background evaluation (TBR) obtained in these summation images does not enable reliable differentiation between low and high grade glioma (LGG and HGG), which, however, can be achieved by dynamicF-FET-PET. We investigated the accuracy of tumour grading using TBR values at different earlier time points after tracer injection. Methods: Three hundred and fourteen patients with histologically proven primary diagnosis of glioma (131 LGG, 183 HGG) who had undergone 40-min dynamic F-FET-PET scans were retrospectively evaluated. TBR was assessed in the standard 20-40 min summation images, as well as in summation images from 0-10 min, 5-15 min, 5-20 min, and 15-30 min p.i., and kinetic analysis was performed. TBR values and kinetic analysis were correlated with histological classification. ROC analyses were performed for each time frame and sensitivity, specificity, and accuracy were assessed. Results: TBR values in the earlier summation images were significantly better for tumour grading ( P < 0.001) when compared to standard 20-40 min scans, with best results for the early 5-15 min scan. This was due to higher TBR in the HGG (3.9 vs. 3.3; p < 0.001), while TBR remained nearly stable in the LGG (2.2 vs. 2.1). Overall, accuracy increased from 70 % in the 20-40 min analysis to 77 % in the 5-15 min images, but did not reach the accuracy of dynamic analysis (80 %). Conclusions: Early TBR assessment (5-15 min p.i.) is more accurate for the differentiation between LGG and HGG than the standard static scan (20-40 min p.i.) mainly caused by the characteristic high F-FET uptake of HGG in the initial phase. Therefore, when dynamic F-FET-PET cannot be performed, early TBR assessment can be considered as an alternative for tumour grading. [ABSTRACT FROM AUTHOR]

Introduction: This study was conducted to compare the association of Gaussian and non-Gaussian magnetic resonance imaging (MRI)-derived parameters with histologic grade and MIB-1 (Ki-67 labeling) index (MI) in brain glioma. Methods: Sixty-five patients with pathologically confirmed glioma, who underwent diffusion-weighted MRI with 2 b values (0, 1000 s/mm) and 22 b values (≤5000 s/mm), respectively, were divided into three groups of grade II ( n = 35), grade III ( n = 8), and grade IV ( n = 22). Comparisons by two groups were made for apparent diffusion coefficient (ADC), slow diffusion coefficient (Dslow), distributed diffusion coefficient (DDC), and heterogeneity index α. Analyses of receiver operating characteristic (ROC) curve were performed to maximize the area under the curve (AUC) for differentiating grade III + IV (high-grade glioma, HGG) from grade II (low-grade glioma, LGG) and grade IV (glioblastoma multiforme, GBM) from grade II + III (other grade glioma, OGG). Correlations with MI were analyzed for the MRI parameters. Results: On tumor regions, the values of ADC, Dslow, DDC, and α were significantly higher in grade II [(1.37 ± 0.29, 0.70 ± 0.11, 1.39 ± 0.34) (×10 mm/s) and 0.88 ± 0.05, respectively] than in grade III [(0.99 ± 0.13, 0.55 ± 0.07, 1.04 ± 0.20) (×10 mm/s) and 0.80 ± 0.03, respectively] and grade IV [(1.03 ± 0.14, 0.50 ± 0.05, 1.02 ± 0.16) (×10 mm/s) and 0.76 ± 0.04, respectively] (all P < 0.001). The parameter α showed the highest AUCs of 0.950 and 0.922 in discriminating HGG from LGG and GBM from OGG, respectively. Significant correlations with histologic grade and MI were observed for the MRI parameters. Conclusion: The non-Gaussian MRI-derived parameters α and Dslow are superior to ADC in glioma grading, which are comparable with ADC as reliable biomarkers in noninvasively predicting the proliferation level of glioma malignancy. [ABSTRACT FROM AUTHOR]

Background and purpose Gliomas are very heterogeneous tumors, glioma grading is currently based on the histologic assessment. With noninvasive measurement of vascular permeability by CT Perfusion (CTP) multiple perfusion parameters which can be obtained with a single acquisition the aim of this study was to find the most sensitive and specific CTP parameters and their cutoffs that can be used to differentiate between low and high grade brain gliomas. Material and methods Twenty-five patients were included in this study divided into two groups: group A includes 15 patients with high grade glioma, group B includes 10 patients with low grade glioma; CTP was done for all patients, perfusion values of tumors were then calculated, and statistical analysis was done using IBM SPSS. Results A statistically highly significant difference was found between the two groups regarding the BF, BV and PS with P values of 0.0003, 0.00026 and 0.0009 respectively. The two groups were found to be self-discriminated by BV and PS with sensitivity of 95% and 86% respectively. Conclusion CTP shows high sensitivity in terms of differentiation between high and low grade adult gliomas. [ABSTRACT FROM AUTHOR]

• Gliomas grading is performed with MRI and is important for treatment and prognosis. • A robust network to handle missing MRI modalities is proposed. • Our network outperformed existing methods. Gliomas are graded using multimodal magnetic resonance imaging, which provides important information for treatment and prognosis. When modalities are missing, the grading is degraded. We propose a robust brain tumor grading model that can handle missing modalities. Our method was developed and tested on Brain Tumor Segmentation Challenge 2017 dataset (n = 285) via nested five-fold cross-validation. Our method adopts adversarial learning to generate the features of missing modalities relative to the features obtained from a full set of modalities in the latent space. An attention-based fusion block across modalities fuses the features of each available modality into a shared representation. Our method's results are compared to those of two other models where 15 missing-modality scenarios are explicitly considered and a joint training approach with random dropouts is used. Our method outperforms the two competing methods in classifying high-grade gliomas (HGGs) and low-grade gliomas (LGGs), achieving an area under the curve of 87.76% on average for all missing-modality scenarios. The activation maps derived with our method confirm that it focuses on the enhancing portion of the tumor in HGGs and on the edema and non-enhancing portions of the tumor in LGGs, which is consistent with prior expertise. An ablation study shows the added benefits of a fusion block and adversarial learning for handling missing modalities. Our method shows robust grading of gliomas in all cases of missing modalities. Our proposed network might have positive implications in glioma care by learning features robust to missing modalities. [ABSTRACT FROM AUTHOR]