Regions Of Peritumoral Edema Research Articles

Deep learning models for rapid discrimination of high-grade gliomas from solitary brain metastases using multi-plane T1-weighted contrast-enhanced (T1CE) images.

High-grade gliomas (HGG) and solitary brain metastases (SBM) are two common types of brain tumors in middle-aged and elderly patients. HGG and SBM display a high degree of similarity on magnetic resonance imaging (MRI) images. Consequently, differential diagnosis using preoperative MRI remains challenging. This study developed deep learning models that used pre-operative T1-weighted contrast-enhanced (T1CE) MRI images to differentiate between HGG and SBM before surgery. By comparing various convolutional neural network models using T1CE image data from The First Medical Center of the Chinese PLA General Hospital and The Second People's Hospital of Yibin (Data collection for this study spanned from January 2016 to December 2023), it was confirmed that the GoogLeNet model exhibited the highest discriminative performance. Additionally, we evaluated the individual impact of the tumoral core and peritumoral edema regions on the network's predictive performance. Finally, we adopted a slice-based voting method to assess the accuracy of the validation dataset and evaluated patient prediction performance on an additional test dataset. The GoogLeNet model, in a five-fold cross-validation using multi-plane T1CE slices (axial, coronal, and sagittal) from 180 patients, achieved an average patient accuracy of 92.78%, a sensitivity of 95.56%, and a specificity of 90.00%. Moreover, on an external test set of 29 patients, the model achieved an accuracy of 89.66%, a sensitivity of 90.91%, and a specificity of 83.33%, with an area under the curve of 0.939 [95% confidence interval (CI): 0.842-1.000]. GoogLeNet performed better than previous methods at differentiating HGG from SBM, even for core and peritumoral edema in both. HGG and SBM could be fast screened using this end-to-end approach, improving workflow for both tumor treatments.

IMG-13. PREDICTION OF PEDIATRIC MEDULLOBLASTOMA SUBGROUPS USING CLINICO-RADIOMIC ANALYSIS

Abstract BACKGROUND Pediatric medulloblastoma is an aggressive brain tumor and tailored treatment has the potential to lead to better patient outcomes. The WHO 2021 classification of medulloblastoma involves an integrated diagnosis that incorporates genetically defined characteristics, including molecular subgroups (WNT-activated, SHH-activated TP53 wildtype, SHH-activated TP53-mutant, and non-WNT/non-SHH). Clinically acquired radiology (MRI) imaging characteristics could provide a non-invasive, pre-treatment biomarker for such subgroups faster than standard methylation-based turn-around times, and for use in countries where methylation is not available. METHODS Herein, we utilize a multi-institutional dataset of multi-parametric, clinical MRIs of pediatric medulloblastoma patients from the Children’s Brain Tumor Network (median age = 8.9 years) to assess the predictive value of patient-level radiomic and clinical factors. 89 subjects with treatment-naïve T1w/T1w contrast-enhanced/T2w/FLAIR images and molecular subgroups (derived from methylation profiling or RNA sequencing) were included (SHH=18, WNT=10, non-SHH/WNT=61). Radiomic features were extracted from a radiologist-defined volumetric (3D) segmentation for each subject separately (including solid tumor, cystic, and peritumoral edema regions). Clinical variables included sex, age at diagnosis, and metastatic disease. 20 top performing features were selected based on an ANOVA between features/classes and subsequently used to train and evaluate three separate classification models based on radiomic, clinical, or clinico-radiomic features (Linear SVM; leave-one-subject-out cross-validation). RESULTS The combined clinico-radiomic model had the top performance (AUC=0.83) followed by radiomic (0.78) and clinical (0.5) for predicting SHH/WNT vs. non-SHH/WNT groups. Predictive radiomic features included intensity-based statistics (T1w, T2w, T1w-CE), texture (T2w), and morphological shape characteristics. CONCLUSIONS Our findings show potential for early prediction of molecular subgroups using baseline imaging that could lend to faster decision-making in patient treatment planning. Future work aims to evaluate the inclusion of histological characteristics for a full integrated diagnostic approach as well as assessment of prognostic value of the combined multi-omic features.

Intratumoral and Peritumoral Edema Radiomics Based on Fat-Suppressed T2- Weighted Imaging for Preoperative Prediction of Triple-Negative Breast Cancer.

Our aim was to explore the feasibility of using radiomics data derived from intratumoral and peritumoral edema on fat-suppressed T2-weighted imaging (T2 FS) to distinguish triple-negative breast cancer (TNBC) from non-triple-negative breast cancer (non-TNBC). This retrospective study enrolled 174 breast cancer patients. According to the MRI examination time, patients before 2021 were divided into training (n = 119) or internal test (n = 30) cohorts at a ratio of 8:2. Patients from 2022 were included in the external test cohort (n = 25). Four regions of interest for each lesion were defined: intratumoral regions, peritumoral edema regions, regions with a combination of intratumoral and peritumoral edema, and regions with a combination of intratumoral and 5-mm peritumoral. Four radiomic signatures were built using the least absolute shrinkage and selection operator (LASSO) method after selecting features. Furthermore, a radio mic-radiological model was constructed using a combination of intratumoral and peritumoral edema regions along with clinical-radiologic features. Area under the receiver operating characteristic curve (AUC) calculations, decision curve analysis, and calibration curve analysis were performed to assess the performance of each model. The radiomic-radiological model showed the highest AUC values of 0.906 (0.788-1.000) and 82.5 (0.622-0.947) in both the internal and external test sets, respectively. The radiology-radiomic model exhibited excellent predictive performance, as evidenced by the calibration curves and decision curve analysis. The ensemble model based on T2 FS-based radiomic features of intratumoral and peritumoral edema, along with radiological factors, performed better in distinguishing TNBC from non-TNBC than a single model. We explored the possibility of developing explainable models to support the clinical decision-making process.

MRI-based intratumoral and peritumoral radiomics for preoperative prediction of glioma grade: a multicenter study.

Accurate preoperative prediction of glioma is crucial for developing individualized treatment decisions and assessing prognosis. In this study, we aimed to establish and evaluate the value of integrated models by incorporating the intratumoral and peritumoral features from conventional MRI and clinical characteristics in the prediction of glioma grade. A total of 213 glioma patients from two centers were included in the retrospective analysis, among which, 132 patients were classified as the training cohort and internal validation set, and the remaining 81 patients were zoned as the independent external testing cohort. A total of 7728 features were extracted from MRI sequences and various volumes of interest (VOIs). After feature selection, 30 radiomic models depended on five sets of machine learning classifiers, different MRI sequences, and four different combinations of predictive feature sources, including features from the intratumoral region only, features from the peritumoral edema region only, features from the fusion area including intratumoral and peritumoral edema region (VOI-fusion), and features from the intratumoral region with the addition of features from peritumoral edema region (feature-fusion), were established to select the optimal model. A nomogram based on the clinical parameter and optimal radiomic model was constructed for predicting glioma grade in clinical practice. The intratumoral radiomic models based on contrast-enhanced T1-weighted and T2-flair sequences outperformed those based on a single MRI sequence. Moreover, the internal validation and independent external test underscored that the XGBoost machine learning classifier, incorporating features extracted from VOI-fusion, showed superior predictive efficiency in differentiating between low-grade gliomas (LGG) and high-grade gliomas (HGG), with an AUC of 0.805 in the external test. The radiomic models of VOI-fusion yielded higher prediction efficiency than those of feature-fusion. Additionally, the developed nomogram presented an optimal predictive efficacy with an AUC of 0.825 in the testing cohort. This study systematically investigated the effect of intratumoral and peritumoral radiomics to predict glioma grading with conventional MRI. The optimal model was the XGBoost classifier coupled radiomic model based on VOI-fusion. The radiomic models that depended on VOI-fusion outperformed those that depended on feature-fusion, suggesting that peritumoral features should be rationally utilized in radiomic studies.

Deep learning and radiomics-based approach to meningioma grading: exploring the potential value of peritumoral edema regions

Objective. To address the challenge of meningioma grading, this study aims to investigate the potential value of peritumoral edema (PTE) regions and proposes a unique approach that integrates radiomics and deep learning techniques. Approach. The primary focus is on developing a transfer learning-based meningioma feature extraction model (MFEM) that leverages both vision transformer (ViT) and convolutional neural network (CNN) architectures. Additionally, the study explores the significance of the PTE region in enhancing the grading process. Main results. The proposed method demonstrates excellent grading accuracy and robustness on a dataset of 98 meningioma patients. It achieves an accuracy of 92.86%, precision of 93.44%, sensitivity of 95%, and specificity of 89.47%. Significance. This study provides valuable insights into preoperative meningioma grading by introducing an innovative method that combines radiomics and deep learning techniques. The approach not only enhances accuracy but also reduces observer subjectivity, thereby contributing to improved clinical decision-making processes.

High-performance presurgical differentiation of glioblastoma and metastasis by means of multiparametric neurite orientation dispersion and density imaging (NODDI) radiomics

ObjectivesTo evaluate the performance of multiparametric neurite orientation dispersion and density imaging (NODDI) radiomics in distinguishing between glioblastoma (Gb) and solitary brain metastasis (SBM).Materials and methodsIn this retrospective study, NODDI images were curated from 109 patients with Gb (n = 57) or SBM (n = 52). Automatically segmented multiple volumes of interest (VOIs) encompassed the main tumor regions, including necrosis, solid tumor, and peritumoral edema. Radiomics features were extracted for each main tumor region, using three NODDI parameter maps. Radiomics models were developed based on these three NODDI parameter maps and their amalgamation to differentiate between Gb and SBM. Additionally, radiomics models were constructed based on morphological magnetic resonance imaging (MRI) and diffusion imaging (diffusion-weighted imaging [DWI]; diffusion tensor imaging [DTI]) for performance comparison.ResultsThe validation dataset results revealed that the performance of a single NODDI parameter map model was inferior to that of the combined NODDI model. In the necrotic regions, the combined NODDI radiomics model exhibited less than ideal discriminative capabilities (area under the receiver operating characteristic curve [AUC] = 0.701). For peritumoral edema regions, the combined NODDI radiomics model achieved a moderate level of discrimination (AUC = 0.820). Within the solid tumor regions, the combined NODDI radiomics model demonstrated superior performance (AUC = 0.904), surpassing the models of other VOIs. The comparison results demonstrated that the NODDI model was better than the DWI and DTI models, while those of the morphological MRI and NODDI models were similar.ConclusionThe NODDI radiomics model showed promising performance for preoperative discrimination between Gb and SBM.Clinical relevance statementThe NODDI radiomics model showed promising performance for preoperative discrimination between Gb and SBM, and radiomics features can be incorporated into the multidimensional phenotypic features that describe tumor heterogeneity.Key Points• The neurite orientation dispersion and density imaging (NODDI) radiomics model showed promising performance for preoperative discrimination between glioblastoma and solitary brain metastasis.• Compared with other tumor volumes of interest, the NODDI radiomics model based on solid tumor regions performed best in distinguishing the two types of tumors.• The performance of the single-parameter NODDI model was inferior to that of the combined-parameter NODDI model.

Diagnostic performance of 3.0 Tesla diffusion tensor imaging in the assessment of brain stem glioma grading.

This study aimed to investigate the role of 3 Tesla Dif-fusion tensor imaging (DTI) in the assessment of brainstem glioma (BSG) grading. The study comprised 22 patients, including pathology-proven 6 brainstem low-grade gliomas (BS-LGG) and 16 brainstem high-grade gliomas (BS-HGG). Characteristics including age, gender, fractional anisotropy (FA), mean diffusivity (MD) of the tumor, peritumoral region, and the ratio of tumor FA to parenchymal FA, as well as tumor MD to parenchymal MD (rFA and rMD), were compared using Mann-Whitney U test, Shapiro-Wilk test, and Chi-square test. Receiver operating characteristic (ROC) curve analysis was used in the study to determine cut-off values and diagnostic values for grading brainstem gliomas (BSG) using diffusion tensor imaging (DTI). Our study revealed no significant difference in age and gender between the BS-LGG and BS-HGG groups (p>0.05). Fractional anisotropy (FA) indices on DTI MRI were found to be highly valuable in grading BSG, with an area under the curve (AUC) of 0.958 - 0.979 when using cut-off values of tFA, pFA, rtFA, and rpFA at 0.318, 0.378, 0.424, and 0.517, respectively. Particularly, rtFA demonstrated the hi-ghest diagnostic value with a sensitivity (Se) of 100%, specificity (Sp) of 93.8%, and AUC of 0.079. Conversely, the indices of tumor mean diffusivity (tMD), peritumoral edema region mean diffusivity (pMD), rtMD, and rpMD showed no diagnostic value in grading BSG. The fractional anisotropy (FA) value on DTI between the tumor region and normal brain parenchyma holds significant value in diagnosing brainstem gliomas (BSG) grading, thereby playing a crucial role in treatment planning and predicting outcomes for patients with brainstem gliomas.

Diagnostic Value of Apparent Diffusion Coefficient Statistics in Differentiating the Glioblastoma Tumors with an Oligodendroglioma and Glioblastomas Without Such Moiety

Background: Differentiating types of brain tumors using magnetic resonance imaging (MRI) as a non-invasive element is essential. Objectives: The study aimed at evaluating the difference in the first-order histogram (FOH) features obtained from apparent diffusion coefficients (ADCs) between glioblastoma (GBM) tumors with or without oligodendroglial (ODG) components. Methods: MRIs were taken before surgical resection of the brain tumors. The brain lesions were analyzed and ADC maps were plotted to calculate FOH. Brain lesions were surgically removed, diagnosed using pathological methods, and categorized as GBM with an ODG (GBM-ODG) or without ODG. A comparison of ADC values was performed between GBM-ODG and GBM. Receiver-operator characteristic (ROC) curves were used to determine the cut-off sensitivity and specificity values. Results: In the GBM group, 26 patients were included; 20 cases were male and 6 were female; the mean ± SD of age was 54.1 ± 1.2 and 55.2 ± 1.6 years, respectively. In the GBM-ODG group, 31 patients were included; 22 cases were male and 9 cases were female; the mean ± SD of 52.1 ± 0.8 and 55.3 ± 1.5 years, respectively. The results of peritumoral edema had several significant FOHs. In the MRIs of tumor and T1W groups, the third moment and minimum-ADC had significant results, respectively. Cut-off values of the FOHs were statistically significant outcomes in the peritumoral edema region when compared between GBM-ODG and GBM categories: Median > 1.3516, normal mean > 0.6671, third moment ≤ -0.0001, 25 percentile > 0.5929, 75 percentile > 0.7336, and 95 percentile > 0.8542. The highest sensitivity and specificity values for these FOHs were higher than 80% and sometimes near 99.9%. Conclusions: The results showed that ADC values from peritumoral edema regions differ between these two types of tumors. Then, an MRI from the edematous areas is suggested when evaluating the patient's status and deciding to make interventions.

Distinguishing Tumor Cell Infiltration and Vasogenic Edema in the Peritumoral Region of Glioblastoma at the Voxel Level via Conventional MRI Sequences

The peritumoral region of glioblastoma (GBM) is composed of infiltrating tumor cells and vasogenic edema, which are difficult to distinguish manually on MRI. To distinguish tumor cell infiltration and vasogenic edema in GBM peritumoral regions, it is crucial to develop a method that is precise, effective, and widely applicable. We retrieved the image characteristics of 379,730 voxels (marker of tumor infiltration) from 28 non-enhanced gliomas and 365,262 voxels (marker of edema) from the peritumoral edema region of 14 meningiomas on conventional MRI sequences (T1-weighted image, the contrast-enhancing T1-weighted image, the T2-weighted image, the T2-fluid attenuated inversion recovery image, and the apparent diffusion coefficient map). Using the SVM classifier, a model for predicting tumor cell infiltration and vasogenic edema at the voxel level was developed. The accuracy of the model's predictions was then evaluated using 15 GBM patients who underwent stereotactic biopsies. The area under the curve (AUC), accuracy, sensitivity, and specificity of the prediction model were 0.93, 0.84, 0.83, and 0.85 in the training set, and 0.90, 0.82, 0.83, and 0.83 in the test set (704,992 voxels), respectively. The pathology verification of 28 biopsy points with an accuracy of 0.79. At the voxel level, it seems possible to forecast tumor cell infiltration and vasogenic edema in the peritumoral region of GBM based on conventional MRI sequences.

Auto-Segmentation and Classification of Glioma Tumors with the Goals of Treatment Response Assessment Using Deep Learning Based on Magnetic Resonance Imaging.

Glioma is the most common primary intracranial neoplasm in adults. Radiotherapy is a treatment approach in glioma patients, and Magnetic Resonance Imaging (MRI) is a beneficial diagnostic tool in treatment planning. Treatment response assessment in glioma patients is usually based on the Response Assessment in Neuro Oncology (RANO) criteria. The limitation of assessment based on RANO is two-dimensional (2D) manual measurements. Deep learning (DL) has great potential in neuro-oncology to improve the accuracy of response assessment. In the current research, firstly, the BraTS 2018 Challenge dataset included 210 HGG and 75 LGGwere applied to train a designed U-Net network for automatic tumor and intra-tumoral segmentation, followed by training of the designed classifier with transfer learning for determining grading HGG and LGG. Then, designed networks were employed for the segmentation and classification of local MRI images of 49 glioma patients preand post-radiotherapy. The results of tumor segmentation and its intra-tumoral regions were utilized to determine the volume of different regions and treatment response assessment. Treatment response assessment demonstrated that radiotherapy is effective on the whole tumor and enhancing region with p-value ≤ 0.05 with a 95% confidence level, while it did not affect necrosis and peri-tumoral edema regions. This work demonstrated the potential of using deep learning in MRI images to provide a beneficial tool in the automated treatment response assessment so that the patient can obtain the best treatment.

Directionally encoded color track density imaging in brain tumor patients: A potential application to neuro-oncology surgical planning

BackgroundDiffusion magnetic resonance imaging white matter tractography, an increasingly popular preoperative planning modality used for pre-surgical planning in brain tumor patients, is employed with the goal of maximizing tumor resection while sparing postoperative neurological function. Clinical translation of white matter tractography has been limited by several shortcomings of standard diffusion tensor imaging (DTI), including poor modeling of fibers crossing through regions of peritumoral edema and low spatial resolution for typical clinical diffusion MRI (dMRI) sequences. Track density imaging (TDI) is a post-tractography technique that uses the number of tractography streamlines and their long-range continuity to map the white matter connections of the brain with enhanced image resolution relative to the acquired dMRI data, potentially offering improved white matter visualization in patients with brain tumors. The aim of this study was to assess the utility of TDI-based white matter maps in a neurosurgical planning context compared to the current clinical standard of DTI-based white matter maps. MethodsFourteen consecutive brain tumor patients from a single institution were retrospectively selected for the study. Each patient underwent 3-Tesla dMRI scanning with 30 gradient directions and a b-value of 1000 s/mm2. For each patient, two directionally encoded color (DEC) maps were produced as follows. DTI-based DEC-fractional anisotropy maps (DEC-FA) were generated on the scanner, while DEC-track density images (DEC-TDI) were generated using constrained spherical deconvolution based tractography. The potential clinical utility of each map was assessed by five practicing neurosurgeons, who rated the maps according to four clinical utility statements regarding different clinical aspects of pre-surgical planning. The neurosurgeons rated each map according to their agreement with four clinical utility statements regarding if the map 1 identified clinically relevant tracts, (2) helped establish a goal resection margin, (3) influenced a planned surgical route, and (4) was useful overall. Cumulative link mixed effect modeling and analysis of variance were performed to test the primary effect of map type (DEC-TDI vs. DEC-FA) on rater score. Pairwise comparisons using estimated marginal means were then calculated to determine the magnitude and directionality of differences in rater scores by map type. ResultsA majority of rater responses agreed with the four clinical utility statements, indicating that neurosurgeons found both DEC maps to be useful. Across all four investigated clinical utility statements, the DEC map type significantly influenced rater score. Rater scores were significantly higher for DEC-TDI maps compared to DEC-FA maps. The largest effect size in rater scores in favor of DEC-TDI maps was observed for clinical utility statement 2, which assessed establishing a goal resection margin. ConclusionWe observed a significant neurosurgeon preference for DEC-TDI maps, indicating their potential utility for neurosurgical planning.

Preoperative MRI for postoperative seizure prediction: a radiomics study of dysembryoplastic neuroepithelial tumor and a systematic review.

In this systematic review the authors aimed to evaluate the effectiveness and superiority of radiomics in detecting tiny epilepsy lesions and to conduct original research in the use of radiomics for preliminary prediction of postoperative seizures in patients with dysembryoplastic neuroepithelial tumor (DNET). The PubMed and Web of Science databases were searched from the earliest record, January 1, 2018, to December 29, 2021, for reports of the detection of epilepsy using radiomics, and the resulting articles were carefully checked according to the PRISMA 2020 guidelines. The authors then conducted original research by evaluating MR images in 18 patients, who were then separated into two groups, the epilepsy recurrence group (ERG) and the epilepsy nonrecurrence group. The tumor region and the edema region were segmented manually by 3D Slicer. The radiomics data were extracted from MR images by using "Slicer Radiomics" running on Mac OS X. Tumor regions were observed with T1-weighted imaging, and edema with FLAIR imaging. Radiomics features with significant differences were selected through comparison according to epilepsy relapses performed with the Mann-Whitney U-test. The edema and tumor regions were also compared within groups to identify their distinctive features. Radiomics features were tested to verify their ability to predict recurrence epilepsy by receiver operating characteristic curve. This systematic review located 9 original articles related to epilepsy and radiomics published from 2018 to 2021. The reported studies demonstrated that radiomics is useful for detecting tiny epilepsy lesions. Among the radiomics features used, the predictive ability of the area under the curve was more than 0.8. The heterogeneity of the peritumoral edema region was found to be higher in the ERG. Satellite lesions in the peritumoral edema region of DNET patients may cause epilepsy recurrence, and radiomics is an emerging method to detect and evaluate these epilepsy-associated lesions.

Logistic Regression-Based Model Is More Efficient Than U-Net Model for Reliable Whole Brain Magnetic Resonance Imaging Segmentation.

Automated whole brain segmentation from magnetic resonance images is of great interest for the development of clinically relevant volumetric markers for various neurological diseases. Although deep learning methods have demonstrated remarkable potential in this area, they may perform poorly in nonoptimal conditions, such as limited training data availability. Manual whole brain segmentation is an incredibly tedious process, so minimizing the data set size required for training segmentation algorithms may be of wide interest. The purpose of this study was to compare the performance of the prototypical deep learning segmentation architecture (U-Net) with a previously published atlas-free traditional machine learning method, Classification using Derivative-based Features (C-DEF) for whole brain segmentation, in the setting of limited training data. C-DEF and U-Net models were evaluated after training on manually curated data from 5, 10, and 15 participants in 2 research cohorts: (1) people living with clinically diagnosed HIV infection and (2) relapsing-remitting multiple sclerosis, each acquired at separate institutions, and between 5 and 295 participants' data using a large, publicly available, and annotated data set of glioblastoma and lower grade glioma (brain tumor segmentation). Statistics was performed on the Dice similarity coefficient using repeated-measures analysis of variance and Dunnett-Hsu pairwise comparison. C-DEF produced better segmentation than U-Net in lesion (29.2%-38.9%) and cerebrospinal fluid (5.3%-11.9%) classes when trained with data from 15 or fewer participants. Unlike C-DEF, U-Net showed significant improvement when increasing the size of the training data (24%-30% higher than baseline). In the brain tumor segmentation data set, C-DEF produced equivalent or better segmentations than U-Net for enhancing tumor and peritumoral edema regions across all training data sizes explored. However, U-Net was more effective than C-DEF for segmentation of necrotic/non-enhancing tumor when trained on 10 or more participants, probably because of the inconsistent signal intensity of the tissue class. These results demonstrate that classical machine learning methods can produce more accurate brain segmentation than the far more complex deep learning methods when only small or moderate amounts of training data are available (n ≤ 15). The magnitude of this advantage varies by tissue and cohort, while U-Net may be preferable for deep gray matter and necrotic/non-enhancing tumor segmentation, particularly with larger training data sets (n ≥ 20). Given that segmentation models often need to be retrained for application to novel imaging protocols or pathology, the bottleneck associated with large-scale manual annotation could be avoided with classical machine learning algorithms, such as C-DEF.

Classification of Gliomas and Germinomas of the Basal Ganglia by Transfer Learning.

BackgroundGerm cell tumors (GCTs) are neoplasms derived from reproductive cells, mostly occurring in children and adolescents at 10 to 19 years of age. Intracranial GCTs are classified histologically into germinomas and non-germinomatous germ cell tumors. Germinomas of the basal ganglia are difficult to distinguish based on symptoms or routine MRI images from gliomas, even for experienced neurosurgeons or radiologists. Meanwhile, intracranial germinoma has a lower incidence rate than glioma in children and adults. Therefore, we established a model based on pre-trained ResNet18 with transfer learning to better identify germinomas of the basal ganglia.MethodsThis retrospective study enrolled 73 patients diagnosed with germinoma or glioma of the basal ganglia. Brain lesions were manually segmented based on both T1C and T2 FLAIR sequences. The T1C sequence was used to build the tumor classification model. A 2D convolutional architecture and transfer learning were implemented. ResNet18 from ImageNet was retrained on the MRI images of our cohort. Class activation mapping was applied for the model visualization.ResultsThe model was trained using five-fold cross-validation, achieving a mean AUC of 0.88. By analyzing the class activation map, we found that the model’s attention was focused on the peri-tumoral edema region of gliomas and tumor bulk for germinomas, indicating that differences in these regions may help discriminate these tumors.ConclusionsThis study showed that the T1C-based transfer learning model could accurately distinguish germinomas from gliomas of the basal ganglia preoperatively.

MRI Radiomic Features of Peritumoral Edema May Predict the Recurrence Sites of Glioblastoma Multiforme

Background: The purpose of this study was to investigate whether there is radiological heterogeneity within peritumoral edema and identify the reproducible radiomic features predictive of the sites of recurrence of glioblastoma multiforme (GBM), which may be of value to optimize patients’ management. Methods: We retrospectively evaluated the clinical information and MR images (T1-weighted and fluid-attenuated inversion recovery sequences) of 22 patients with histologically diagnosed GBM. Kaplan-Meier’s method was used for survival analysis. Peritumoral edema regions were manually segmented by an expert into recurrence and non-recurrence subregions. A set of 94 radiomic features were obtained from both subregions using a three-dimensional slicer. Features with significant differences between the two subregions were identified using the paired t test. Subsequently, the data of two patients from The Cancer Genome Atlas (TCGA) database were used to evaluate the clinical value of the selected features. Findings: Ten features with significant differences between the recurrence and non-recurrence subregions were identified and verified on two individual patients from TCGA database with pathologically confirmed diagnosis of GBM. Interpretation: Our results suggested that heterogeneity does exist in peritumoral edema, indicating that the radiomic features of peritumoral edema from routine MR images can be utilized to predict the sites of GBM recurrence. Our findings may further guide the surgical treatment strategy for GBM. Funding Information: This work was supported by grants from the Natural Science Fund of Guangdong Province (No. 2017A030313597), “Climbing Program” Special Fund of Guangdong Province (No. pdjh2019b0100, No. pdjh2020b0112) and Southern Medical University (No. LX2016N006, No. KJ20161102, No.201912121004S, No.201912121013，No. S202012121088, No. X202012121354). Declaration of Interests: The authors of this manuscript declare no relationships with any companies, whose products or services may be related to the subject matter of the article. Ethics Approval Statement: Ethics committee approval and patient consent were obtained.

Specific Preoperative Dynamic Contrast-Enhanced MRI Semi-quantitative Markers Can Correlate With Vascularity in Specific Areas of Glioblastoma Tissue and Predict Recurrence.

Background: Glioblastoma is one of the most aggressive tumours despite all advanced therapies. We aimed to investigate the correlation between qualitative markers of dynamic contrast-enhanced magnetic resonance imaging and vascularity in different tumour regions and elucidate their potential in predicting recurrence.Methods: Radiological markers of vascularity as wash-in rate, washout rate, and capillary time to peak in different single tumour regions were extracted for all glioblastoma patients before being surgically resected using preoperative dynamic contrast-enhanced MRI (DCE-MRI). Tissue samples were obtained from different intratumoral regions and peritumoral oedema and evaluated for the vascular endothelial growth factor (VEGF).Results: Two hundred sixty individuals were included in the final analysis, with 180 dead ones and 80 survivors. Radio- and chemo-therapy were received by all surviving patients and 77.8% (n= 140) of the dead ones. The mean time to peak, in seconds, was longest at the peritumoral oedema region (71.7±23.5), followed by the tumour's necrotic centre (50.0±28.5) and its periphery (2.9±1.8). The expression of VEGF at the peritumoral edema region was inversely correlated to the washout rate at the periphery (r= -0.66; P-value= 0.014) and positively correlated to peritumoral TTP (r= 0.94; P-value< 0.001).Conclusion: Using DCE-MRI, VEGF expression may be used as a non-invasive marker to estimate tumour grade for clinical diagnosis and treatment. Moreover, the risk of glioblastoma recurrence could be determined by evaluating the washout rate at the tumour's periphery. Further large-scale studies are needed to validate the results and to have concrete evidence.

A Multiparametric MRI-Based Radiomics Analysis to Efficiently Classify Tumor Subregions of Glioblastoma: A Pilot Study in Machine Learning

Glioblastoma multiforme (GBM) carries a poor prognosis and usually presents with heterogenous regions of a necrotic core, solid part, peritumoral tissue, and peritumoral edema. Accurate demarcation on magnetic resonance imaging (MRI) between the active tumor region and perifocal edematous extension is essential for planning stereotactic biopsy, GBM resection, and radiotherapy. We established a set of radiomics features to efficiently classify patients with GBM and retrieved cerebral multiparametric MRI, including contrast-enhanced T1-weighted (T1-CE), T2-weighted, T2-weighted fluid-attenuated inversion recovery, and apparent diffusion coefficient images from local patients with GBM. A total of 1316 features on the raw MR images were selected for each annotated area. A leave-one-out cross-validation was performed on the whole dataset, the different machine learning and deep learning techniques tested; random forest achieved the best performance (average accuracy: 93.6% necrosis, 90.4% solid part, 95.8% peritumoral tissue, and 90.4% peritumoral edema). The features from the enhancing tumor and the tumor shape elongation of peritumoral edema region for high-risk groups from T1-CE. The multiparametric MRI-based radiomics model showed the efficient classification of tumor subregions of GBM and suggests that prognostic radiomic features from a routine MRI exam may also be significantly associated with key biological processes that affect the response to chemotherapy in GBM.

Clinical Significance of Fractional Anisotropy Measured in Peritumoral Edema as a Biomarker of Overall Survival in Glioblastoma: Evidence Using Correspondence Analysis.

Fractional anisotropy (FA), a diffusion tensor image (DTI) derived biomarker is related to invasion, infiltration, and extension of glioblastoma (GB). We aimed to evaluate FA values and their association with intervals of overall survival (OS). Retrospective study conducted in 36 patients with GB included 23 (63.9%) males, 46 ± 14 y; and 13 (36.1%) females, 53 ± 13; followed up for 36 months. We measured FA at edema, enhancing rim, and necrosis. We created two categorical variables using levels of FA and intervals of OS to evaluate their relationships. Kaplan-Meier method and correspondence analysis evaluated the association between OS (grouped in 7 six-month intervals) and FA measurements. Median FA values were higher in healthy brain regions (0.351), followed by peritumoral edema (0.190), enhancing ring (0.116), and necrosis (0.071). Pair-wise comparisons among tumor regions showed a significant difference, P < 0.001. The median OS for all patients was 19.3 months; variations in the OS curves among subgroups was significant χ2 (3) = 8.48, P = 0.037. Correspondence analysis showed a significant association between FA values in the edema region and the survival intervals χ2 (18) = 30.996, P = 0.029. Alternative multivariate assessment using correspondence analysis might supplement the traditional survival analysis in patients with GB. A close follow-up of the variability of FA in the peritumoral edema region is predictive of the OS within specific six-month interval subgroup. Further studies should focus on predictive models combining surgical and DTI biomarkers.

Quantitative histopathologic assessment of perfusion MRI as a marker of glioblastoma cell infiltration in and beyond the peritumoral edema region.

Conventional MRI fails to detect regions of glioblastoma cell infiltration beyond the contrast-enhanced T1 solid tumor region, with infiltrating tumor cells often migrating along host blood vessels. To quantitatively and qualitatively analyze the correlation between perfusion MRI signal and tumor cell density in order to assess whether local perfusion perturbation could provide a useful biomarker of glioblastoma cell infiltration. Animal model. Mice bearing orthotopic glioblastoma xenografts generated from a patient-derived glioblastoma cell line. 7T perfusion images acquired using a high signal-to-noise ratio (SNR) multiple boli arterial spin labeling sequence were compared with conventional MRI (T1 /T2 weighted, contrast-enhanced T1 , diffusion-weighted, and apparent diffusion coefficient). Immunohistochemistry sections were stained for human leukocyte antigen (probing human-derived tumor cells). To achieve quantitative MRI-tissue comparison, multiple histological slices cut in the MRI plane were stacked to produce tumor cell density maps acting as a "ground truth." Sensitivity, specificity, accuracy, and Dice similarity indices were calculated and a two-tailed, paired t-test used for statistical analysis. High comparison test results (Dice 0.62-0.72, Accuracy 0.86-0.88, Sensitivity 0.51-0.7, and Specificity 0.92-0.97) indicate a good segmentation for all imaging modalities and highlight the quality of the MRI tissue assessment protocol. Perfusion imaging exhibits higher sensitivity (0.7) than conventional MRI (0.51-0.61). MRI/histology voxel-to-voxel comparison revealed a negative correlation between tumor cell infiltration and perfusion at the tumor margins (P = 0.0004). These results demonstrate the ability of perfusion imaging to probe regions of low tumor cell infiltration while confirming the sensitivity limitations of conventional imaging modalities. The quantitative relationship between tumor cell density and perfusion identified in and beyond the edematous T2 hyperintensity region surrounding macroscopic tumor could be used to detect marginal tumor cell infiltration with greater accuracy. 1 Technical stage: 2 J. Magn. Reson. Imaging 2019;50:529-540.

P044] Imaging biomarker analysis of structural MR images for glioma grading

Purpose Preoperative differentiation between low and high grade gliomas is critical for adapting appropriate treatment strategies. However biopsy based diagnosis presents limitations in assessing tumor diffusive and solid growth patterns, respectively, due to the potential subsampling of tumor region as well as the associated patient risk. The purpose of this study is to investigate the value of imaging biomarker analysis in terms of a 3D texture analysis of T1 pre/post-contrast, T2, and DTI maps in the evaluation of tumor heterogeneity towards glioma grading. Methods In this study a semi-automated segmentation technique, based on T1 pre/post-contrast, T2, and DTI parametric maps, is used for the identification of the tumor core and peritumoral edema regions from 3 Tesla MR Data. 3D Texture analysis was performed on 10 low grade and 10 high grade glioma patients, considering several 3D Histogram, Gray Level Co-Occurrence Matrices (GLCM) and Gray Level Run Length Matrices (GLRLM) derived features. Results Initial evaluation of the proposed method is performed in terms of textural features statistical significant differences between the two tumor groups, with the prospect of implementing an advance classification scheme for identifying more complex correlations. The preliminary results suggest that 3D Texture Analysis findings might be utilized as a surrogate biomarker for the global non-invasive assessment of glioma grading. Conclusions Results demonstrate the potential role of Texture Analysis of 3D tumor imaging appearance in structural MRI, for pre-treatment gliomas grade differentiation.