Non-enhancing Tumor Research Articles

Automatic segmentation of glioblastoma multiform brain tumor in MRI images: Using Deeplabv3+ with pre-trained Resnet18 weights

PurposeTo assess the effectiveness of deep learning algorithms in automated segmentation of magnetic resonance brain images for determining the enhanced tumor, the peri-tumoral edema, the necrotic/ non-enhancing tumor, and Normal tissue volumes. Methods and MaterialsA new deep neural network algorithm, Deep-Net, was developed for semantic segmentation of the glioblastoma tumors in MR images, using the Deeplabv3+ architecture, and the pre-trained Resnet18 initial weights. The MR image Dataset used for training the network was taken from the BraTS 2020 training set, with the ground truth labels for different tumor subregions manually drawn by a group of expert neuroradiologists. In this work, two multi-modal MRI scans, i.e., T1ce and FLAIR of 293 patients with high-grade glioma (HGG), were used for deep network training (Deep-Net). The performance of the network was assessed for different hyper-parameters, to obtain the optimum set of parameters. The similarity scores were used for the evaluation of the optimized network. ResultsAccording to the results of this study, epoch #37 is the optimum epoch giving the best global accuracy (97.53%), and loss function (0.14). The Deep-Net sensitivity in the delineation of the enhanced tumor is more than 90%. ConclusionsThe results indicate that the Deep-Net was able to segment GBM tumors with high accuracy.

Progression Patterns in Non-Contrast-Enhancing Gliomas Support Brain Tumor Responsiveness to Surgical Lesions.

Purpose: The overall benefit of surgical treatments for patients with glioma is undisputed. We have shown preclinically that brain tumor cells form a network that is capable of detecting damage to the tumor, and repair itself. The aim of this study was to determine whether a similar mechanism might contribute to local recurrence in the clinical setting. Methods: We evaluated tumor progression patterns of 24 initially non-contrast-enhancing gliomas that were partially resected or biopsied. We measured the distance between the new contrast enhancement developing over time, and prior surgical lesioning, and evaluated tumor network changes in response to sequential resections by quantifying tumor cells and tumor networks with specific stainings against IDH1-R132H. Results: We found that new contrast enhancement appeared within the residual, non-enhancing tumor mass in 21/24 patients (87.5%). The location of new contrast enhancement within the residual tumor region was non-random; it occurred adjacent to the wall of the resection cavity in 12/21 patients (57.1%). Interestingly, the density of the glioma cell network increased in all patient tumors between initial resection or biopsy and recurrence. In line with the histological and radiological malignization, Ki67 expression increased from initial to final resections in 14/17 cases. Conclusion: The non-random distribution of glioma malignization in patients and unidirectional increase of anatomical tumor networks after surgical procedures provides evidence that surgical lesions, in the presence of residual tumor cells, can stimulate local tumor progression and tumor cell network formation. This argues for the development of intraoperative treatments increasing the benefits from surgical resection by specifically disrupting the mechanisms of local recurrence, particularly tumor cell network functionality.

Edge Constraint and Location Mapping for Liver Tumor Segmentation from Nonenhanced Images.

As there is no contrast enhancement, the liver tumor area in nonenhanced MRI exists with blurred edges and low contrast, which greatly affects the speed and accuracy of liver tumor diagnosis. As a result, precise segmentation of liver tumor from nonenhanced MRI has become an urgent and challenging task. In this paper, we propose an edge constraint and localization mapping segmentation model (ECLMS) to accurately segment liver tumor from nonenhanced MRI. It consists of two parts: localization network and dual-branch segmentation network. We build the localization network, which generates prior coarse masks to provide position mapping for the segmentation network. This part enhances the ability of the model to localize liver tumor in nonenhanced images. We design a dual-branch segmentation network, where the main decoding branch focuses on the feature representation in the core region of the tumor and the edge decoding branch concentrates on capturing the edge information of the tumor. To improve the ability of the model for capturing detailed features, sSE blocks and dense upward connections are introduced into it. We design the bottleneck multiscale module to construct multiscale feature representations using kernels of different sizes while integrating the location mapping of tumor. The ECLMS model is evaluated on a private nonenhanced MRI dataset that comprises 215 different subjects. The model achieves the best Dice coefficient, precision, and accuracy of 90.23%, 92.25%, and 92.39%, correspondingly. The effectiveness of our model is demonstrated by experiment results, and our model reaches superior results in the segmentation task of nonenhanced liver tumor compared to existing segmentation methods.

Imaging Glioblastoma With 18F-Fluciclovine Amino Acid Positron Emission Tomography.

IntroductionConventional methods of imaging brain tumors fail to assess metabolically active tumor regions, which limits their capabilities for tumor detection, localization, and response assessment. Positron emission tomography (PET) with 18F-fluciclovine (fluciclovine) provides regional assessment of amino acid uptake in tumors that could overcome some of the limitations of conventional imaging. However, the biological basis of enhanced fluciclovine uptake is insufficiently characterized in brain tumors, which confounds clinical interpretation and application. This study sought to address this gap by correlating multiple biologic quantities with fluciclovine PET uptake across a range of human glioblastoma xenograft models.MethodsThirty-one rats underwent orthotopic implantations with one of five different human glioblastoma cell lines. After tumors were established, fluciclovine PET and magnetic resonance imaging (MRI) scans were performed. The fluciclovine tumor-to-normal-brain (TN) uptake ratio was used to quantify fluciclovine uptake. MRI scans were used to assess tumor volume and gadolinium enhancement status. Histologic assessments quantified tumor cell proliferation, tumor cell density, and tumor cell amino acid transporters (LAT1 and ASCT2). Multivariate linear regression models related fluciclovine uptake with the other measured quantities.ResultsWithin the multivariate regression, the fluciclovine TN uptake ratio (measured 15 to 35 minutes after fluciclovine injection) was most strongly associated with tumor ASCT2 levels (β=0.64; P=0.001). The fluciclovine TN uptake ratio was also significantly associated with tumor volume (β=0.45; P=0.001) and tumor enhancement status (β=0.40; P=0.01). Tumor cell proliferation, tumor cell density, and LAT1 levels were not significantly associated with fluciclovine uptake in any of the multivariate models. In general, both enhancing and non-enhancing tumors could be visualized on fluciclovine PET images, with the median TN uptake ratio across the five tumor lines being 2.4 (range 1.1 to 8.9).ConclusionsIncreased fluciclovine PET uptake was associated with increased levels of the amino acid transporter ASCT2, suggesting fluciclovine PET may be useful for assessing brain tumor amino acid metabolism. Fluciclovine PET uptake was elevated in both enhancing and non-enhancing tumors but the degree of uptake was greater in larger tumors and tumors with enhancement, indicating these variables could confound fluciclovine metabolic measurements if not accounted for.

Prediction and Visualization of Non-Enhancing Tumor in Glioblastoma via T1w/T2w-Ratio Map.

One of the challenges in glioblastoma (GBM) imaging is to visualize non-enhancing tumor (NET) lesions. The ratio of T1- and T2-weighted images (rT1/T2) is reported as a helpful imaging surrogate of microstructures of the brain. This research study investigated the possibility of using rT1/T2 as a surrogate for the T1- and T2-relaxation time of GBM to visualize NET effectively. The data of thirty-four histologically confirmed GBM patients whose T1-, T2- and contrast-enhanced T1-weighted MRI and 11C-methionine positron emission tomography (Met-PET) were available were collected for analysis. Two of them also underwent MR relaxometry with rT1/T2 reconstructed for all cases. Met-PET was used as ground truth with T2-FLAIR hyperintense lesion, with >1.5 in tumor-to-normal tissue ratio being NET. rT1/T2 values were compared with MR relaxometry and Met-PET. rT1/T2 values significantly correlated with both T1- and T2-relaxation times in a logarithmic manner (p < 0.05 for both cases). The distributions of rT1/T2 from Met-PET high and low T2-FLAIR hyperintense lesions were different and a novel metric named Likeliness of Methionine PET high (LMPH) deriving from rT1/T2 was statistically significant for detecting Met-PET high T2-FLAIR hyperintense lesions (mean AUC = 0.556 ± 0.117; p = 0.01). In conclusion, this research study supported the hypothesis that rT1/T2 could be a promising imaging marker for NET identification.

Fluorescence-guided craniotomy of glioblastoma using panitumumab-IRDye800.

A contrast-enhancing lesion in the left temporal lobe of a 72-year-old woman was biopsied and diagnosed as glioblastoma. Near-infrared (NIR)–labeled epidermal growth factor receptor (EGFR) antibody, panitumumab-IRDye800, was infused 52 hours before craniotomy without pretreatment. Tumor fluorescence was detected through intact dura, and the visual contrast between disease and peritumoral healthy brain was enhanced after tumor exposure. Residual cancerous tissue was identified with strong fluorescence in resection cavity after en bloc tumor removal. Minimal fluorescence remained in the final wound bed, likely from nonenhancing tumor. Fluorescence was heterogeneously distributed at the infiltrative margin in resected tumor pieces imaged ex vivo. Postoperative MRI confirmed gross-total resection.The video can be found here: https://stream.cadmore.media/r10.3171/2021.10.FOCVID21201

Relax and focus on brain tumor segmentation.

In this paper, we present a Deep Convolutional Neural Networks (CNNs) for fully automatic brain tumor segmentation for both high- and low-grade gliomas in MRI images. Unlike normal tissues or organs that usually have a fixed location or shape, brain tumors with different grades have shown great variation in terms of the location, size, structure, and morphological appearance. Moreover, the severe data imbalance exists not only between the brain tumor and non-tumor tissues, but also among the different sub-regions inside brain tumor (e.g., enhancing tumor, necrotic, edema, and non-enhancing tumor). Therefore, we introduce a hybrid model to address the challenges in the multi-modality multi-class brain tumor segmentation task. First, we propose the dynamic focal Dice loss function that is able to focus more on the smaller tumor sub-regions with more complex structures during training, and the learning capacity of the model is dynamically distributed to each class independently based on its training performance in different training stages. Besides, to better recognize the overall structure of the brain tumor and the morphological relationship among different tumor sub-regions, we relax the boundary constraints for the inner tumor regions in coarse-to-fine fashion. Additionally, a symmetric attention branch is proposed to highlight the possible location of the brain tumor from the asymmetric features caused by growth and expansion of the abnormal tissues in the brain. Generally, to balance the learning capacity of the model between spatial details and high-level morphological features, the proposed model relaxes the constraints of the inner boundary and complex details and enforces more attention on the tumor shape, location, and the harder classes of the tumor sub-regions. The proposed model is validated on the publicly available brain tumor dataset from real patients, BRATS 2019. The experimental results reveal that our model improves the overall segmentation performance in comparison with the state-of-the-art methods, with major progress on the recognition of the tumor shape, the structural relationship of tumor sub-regions, and the segmentation of more challenging tumor sub-regions, e.g., the tumor core, and enhancing tumor.

Early-Stage Segmentation and Characterization of Brain Tumor

Gliomas are the most aggressive brain tumors caused by the abnormal growth of brain tissues. The life expectancy of patients diagnosed with gliomas decreases exponentially. Most gliomas are diagnosed in later stages, resulting in imminent death. On average, patients do not survive 14 months after diagnosis. The only way to minimize the impact of this inevitable disease is through early diagnosis. The Magnetic Resonance Imaging (MRI) scans, because of their better tissue contrast, are most frequently used to assess the brain tissues. The manual classification of MRI scans takes a reasonable amount of time to classify brain tumors. Besides this, dealing with MRI scans manually is also cumbersome, thus affects the classification accuracy. To eradicate this problem, researchers have come up with automatic and semi-automatic methods that help in the automation of brain tumor classification task. Although, many techniques have been devised to address this issue, the existing methods still struggle to characterize the enhancing region. This is because of low variance in enhancing region which give poor contrast in MRI scans. In this study, we propose a novel deep learning based method consisting of a series of steps, namely: data pre-processing, patch extraction, patch pre-processing, and a deep learning model with tuned hyper-parameters to classify all types of gliomas with a focus on enhancing region. Our trained model achieved better results for all glioma classes including the enhancing region. The improved performance of our technique can be attributed to several factors. Firstly, the non-local mean filter in the pre-processing step, improved the image detail while removing irrelevant noise. Secondly, the architecture we employ can capture the non-linearity of all classes including the enhancing region. Overall, the segmentation scores achieved on the Dice Similarity Coefficient (DSC) metric for normal, necrosis, edema, enhancing and non-enhancing tumor classes are 0.95, 0.97, 0.91, 0.93, 0.95; respectively.

The impact of different volumetric thresholds to determine progressive disease in patients with recurrent glioblastoma treated with bevacizumab.

BackgroundThe optimal volumetric threshold for determining progressive disease (PD) in recurrent glioblastoma is yet to be determined. We investigated a range of thresholds in association with overall survival (OS).MethodsFirst recurrent glioblastoma patients treated with bevacizumab and/or lomustine were included from the phase II BELOB and phase III EORTC26101 trials. Enhancing and nonenhancing tumor volumes were measured at baseline, first (6 weeks), and second (12 weeks) follow-up. Hazard ratios (HRs) for the appearance of new lesions and several thresholds for tumor volume increase were calculated using cox regression analysis. Results were corrected in a multivariate analysis for well-established prognostic factors.ResultsAt first and second follow-up, 138 and 94 patients respectively, were deemed eligible for analysis of enhancing volumes, while 89 patients were included in the analysis of nonenhancing volumes at first follow-up. New lesions were associated with a significantly worse OS (3.2 versus 11.2 months, HR = 7.03, P < .001). At first follow-up a threshold of enhancing volume increase of ≥20% provided the highest HR (5.55, p = .001. At second follow-up, any increase in enhancing volume (≥0%) provided the highest HR (9.00, p < .001). When measuring nonenhancing volume at first follow-up, only 6 additional patients were scored as PD with the highest HR of ≥25% increase in volume (HR=3.25, p = .008).ConclusionEarly appearing new lesions were associated with poor OS. Lowering the volumetric threshold for PD at both first and second follow-up improved survival prediction. However, the additional number of patients categorized as PD by lowering the threshold was very low. The per-RANO added change in nonenhancing volumes to the analyses was of limited value.

Changes in the Relapse Pattern and Prognosis of Glioblastoma After Approval of First-Line Bevacizumab: A Single-Center Retrospective Study

Controversies exist regarding the aggressive recurrence of glioblastoma after bevacizumab treatment. We analyzed the clinical impact of bevacizumab approval in Japan by evaluating the clinical course and relapse pattern in patients with glioblastoma. We included 100 patients with IDH-wild-type glioblastoma from September 2006 to February 2018 in our institution. The patients were classified into the pre-bevacizumab (n= 51) and post-bevacizumab (n= 49) groups. Overall, progression-free, deterioration-free, and postprogression survivals were compared. We analyzed the relapse pattern of 72 patients, whose radiographic progressions were evaluated. Significant improvement in progression-free (pre-bevacizumab, 7.5 months; post-bevacizumab, 9.9 months; P= 0.0153) and deterioration-free (pre-bevacizumab, 8.5 months; post-bevacizumab, 13.8 months; P= 0.0046) survivals was seen. These survival prolongations were strongly correlated (r: 0.91, P < 0.0001). The nonenhancing tumor pattern was novel in the post-bevacizumab era (5 of 33). The presence of a nonenhancing tumor did not indicate poor postprogression survival (hazard ratio: 0.82 [0.26-2.62], P= 0.7377). The rate of early focal recurrence was significantly lower (P=0.0155) in the post-bevacizumab (4 of 33) than in the pre-bevacizumab (18 of 39) era. There was a significant decrease in early focal recurrence after approval of bevacizumab in patients with unresectable tumors (P= 0.0110). The treatment era was significantly correlated with a decreased rate of early focal recurrence (P= 0.0021, univariate analysis; P= 0.0144, multivariate analysis). Approval of first-line bevacizumab in Japan for unresectable tumors may prevent early progression and clinical deterioration of glioblastoma without worsening the clinical course after relapse.

Effects of Multi-Shell Free Water Correction on Glioma Characterization.

Diffusion MRI is a useful tool to investigate the microstructure of brain tumors. However, the presence of fast diffusing isotropic signals originating from non-restricted edematous fluids, within and surrounding tumors, may obscure estimation of the underlying tissue characteristics, complicating the radiological interpretation and quantitative evaluation of diffusion MRI. A multi-shell regularized free water (FW) elimination model was therefore applied to separate free water from tissue-related diffusion components from the diffusion MRI of 26 treatment-naïve glioma patients. We then investigated the diagnostic value of the derived measures of FW maps as well as FW-corrected tensor-derived maps of fractional anisotropy (FA). Presumed necrotic tumor regions display greater mean and variance of FW content than other parts of the tumor. On average, the area under the receiver operating characteristic (ROC) for the classification of necrotic and enhancing tumor volumes increased by 5% in corrected data compared to non-corrected data. FW elimination shifts the FA distribution in non-enhancing tumor parts toward higher values and significantly increases its entropy (p ≤ 0.003), whereas skewness is decreased (p ≤ 0.004). Kurtosis is significantly decreased (p < 0.001) in high-grade tumors. In conclusion, eliminating FW contributions improved quantitative estimations of FA, which helps to disentangle the cancer heterogeneity.

NI-2 Use of neurite orientation dispersion and density imaging(NODDI)for early distinction between infiltrating tumor and vasogenic edema in non-enhancing lesions with glioblastoma patients

Background: Glioblastoma is a highly infiltrative tumor. In the non-enhancing T2-weighted hyperintense area, differentiating between non-enhancing tumors (NETs) and vasogenic edema is challenging. Neurite orientation dispersion and density imaging (NODDI) is a new diffusion MRI technique that reveals the inhomogeneity of the brain microstructure. The aim of this study is to differentiate between NETs and edema in glioblastomas using NODDI. Methods: Data were collected from 20 patients with glioblastoma as well as three patients with metastasis and two with meningioma (control), who underwent MRI as part of pre-surgical examination. The MRI data included T2- and T1-weighted contrast-enhanced images and NODDI images. Three neurosurgeons manually placed the volume of interest (VOI) on the NETs and edema based on the previous reports. ICVF, ODI, ISOVF, FA, and ADC were calculated for each VOI. Results: Fifteen and 13 VOIs were placed on NETs and edema, respectively. Each parameter was measured and the unpaired t-test revealed a significant difference between NETs and edema (p <0.0001). The ROC curve analysis revealed a large difference in the ADC, FA, and ISOVF between NETs and edema compared to ICVF and ODI. Principal component analysis of the five parameters showed that ADC, ISOVF, and FA contributed to the differentiation between NETs and edema. Multiple logistic regression analysis was performed with the three aforementioned parameters. A predictive formula could be created to discriminate between NETs and edema, following the use of which, the ROC curve revealed an AUC value of 0.8891. Furthermore, this formula was applied to the edematous regions of the images of the negative control group, and the prediction degree of the tumor was well below 0.5, thus enabling differentiation as edema.Conclusions: NODDI may prove to be a useful tool to discriminate between NETs and edema in the non-contrast T2 hyperintensity region of glioblastoma.

NI-14 estimation of property of MRI non-contrast enhanced lesion of Glioblastoma using T1/T2 ratio

Background: Tumor mass of glioblastoma is considered to exist beyond gadolinium-enhancing lesion into T2/FLAIR-high intensity lesions (T2/FL-HIL) on MRI. However, it is challenging to differentiate non-enhancing tumor region (NET) from pure brain edema for T2/FL-HIL. The T1/T2 ratio (rT1/T2) is an MRI metric considered to semi-quantify the tissue relaxation time on MRI. This research tested the hypothesis that rT1/T2 is useful for identifying NET within T2/FL-HIL by comparing it with 11C-methionine positron emission tomography (MET-PET). Method: Forty-six glioblastoma (GBM) patients at Osaka International Cancer Institute and Osaka University Hospital where T1-, T2- and contrast-enhanced T1-weighted MRI and MET-PET were available were included in this study. rT1/T2 maps were obtained after signal corrections were performed, as reported previously. Region-of-interests (ROIs) were defined within T2/FL-HILs beyond the gadolinium-enhanced lesion. MET-PET and rT1/T2 maps were co-registered to the same coordinate system, and the relationship between methionine uptake and rT1/T2 values was examined in a voxel-wise manner.ResultApproximately three million voxels were included for analysis. Lesions with methionine uptake higher than 5.0 on T/N showed 0.7 < rT1/T2 < 0.98. For those with methionine uptake higher than 3.0, rT1/T2 was between 0.70 and 1.04.DiscussionThis report suggested that rT1/T2 represents histological characteristics of the glioblastoma within T2/FL-HIL. It also indicated that rT1/T2 could be a useful biomarker for detecting NET within T2/FL-HIL for glioblastoma.

NI-3 Magnetic resonance relaxometry for tumor cell density imaging for glioma: An exploratory study via 11C-methionine PET and its validation via stereotactic tissue sampling

Objective: While visualization of non-enhancing tumors for glioma is crucial for planning the most appropriate surgical or non-surgical treatment of the disease, current MRI cannot achieve this goal. This study aims to test the hypothesis that quantitative and diffusion MRI can estimate tumor burden with the brain. Materials and Methods: Study 1: Ten patients who have undergone Methionine PET (Met-PET), quantitative MRI (qMRI), and diffusion MRI (DWI) were included for analysis. A cut-off of a tumor-to-normal ratio (T/Nr) 1.5 was set on Met-PET, and the values from qMRI and DWI were compared. Study 2: Seventy-nine stereo-tactically sampled tissues from 22 glioma patients were correlated with Met-PET, qMRI, and DWI measurements regarding tumor cell density. qMRI acquisition: Imaging was performed on either a 1.5 or 3 T MR scanner (Prisma or Aera; Siemens Healthcare, Erlangen, Germany). T1-relaxometry was achieved by first acquiring MP2RAGE images, then converting those images into T1-relaxation time maps. At the same time, T2-relaxometry was achieved by first acquiring multi-echo T2-weighted images and then converting those images into T2-relaxation time maps, with both relaxometries performed via Bayesian inference modeling (Olea Nova+; Canon Medical Systems, Tochigi, Japan). Results: Study 1 revealed that regions of 1850ms < T1-relaxation time < 3200ms and 115ms < T2-relaxation time < 225ms tended to be Met-PET T/Nr > 1.5. DWI was not useful to separate areas between low and high Met-PET. Study 2 showed that regions of 1850ms < T1-relaxation time < 3200ms showed high tumor cell density than other areas (p=0.04). Conclusions: Our results supported the hypothesis that qMRI is useful for predicting the tumor load within the brain among glioma patients. T1-relaxation time was notably useful for this means. On the other hand, ADC measured from DWI was limited for tumor load prediction.

Comparison of machine learning classifiers to predict patient survival and genetics of GBM: towards a standardized model for clinical implementation

Radiomic models have been shown to outperform clinical data for outcome prediction in glioblastoma (GBM). However, clinical implementation is limited by lack of parameters standardization. We aimed to compare nine machine learning classifiers, with different optimization parameters, to predict overall survival (OS), isocitrate dehydrogenase (IDH) mutation, O-6-methylguanine-DNA-methyltransferase (MGMT) promoter methylation, epidermal growth factor receptor (EGFR) VII amplification and Ki-67 expression in GBM patients, based on radiomic features from conventional and advanced MR. 156 adult patients with pathologic diagnosis of GBM were included. Three tumoral regions were analyzed: contrast-enhancing tumor, necrosis and non-enhancing tumor, selected by manual segmentation. Radiomic features were extracted with a custom version of Pyradiomics, and selected through Boruta algorithm. A Grid Search algorithm was applied when computing 4 times K-fold cross validation (K=10) to get the highest mean and lowest spread of accuracy. Once optimal parameters were identified, model performances were assessed in terms of Area Under The Curve-Receiver Operating Characteristics (AUC-ROC). Metaheuristic and ensemble classifiers showed the best performance across tasks. xGB obtained maximum accuracy for OS (74.5%), AB for IDH mutation (88%), MGMT methylation (71,7%), Ki-67 expression (86,6%), and EGFR amplification (81,6%). Best performing features shed light on possible correlations between MR and tumor histology.

AI and High-Grade Glioma for Diagnosis and Outcome Prediction: Do All Machine Learning Models Perform Equally Well?

Radiomic models outperform clinical data for outcome prediction in high-grade gliomas (HGG). However, lack of parameter standardization limits clinical applications. Many machine learning (ML) radiomic models employ single classifiers rather than ensemble learning, which is known to boost performance, and comparative analyses are lacking in the literature. We aimed to compare ML classifiers to predict clinically relevant tasks for HGG: overall survival (OS), isocitrate dehydrogenase (IDH) mutation, O-6-methylguanine-DNA-methyltransferase (MGMT) promoter methylation, epidermal growth factor receptor vIII (EGFR) amplification, and Ki-67 expression, based on radiomic features from conventional and advanced magnetic resonance imaging (MRI). Our objective was to identify the best algorithm for each task. One hundred fifty-six adult patients with pathologic diagnosis of HGG were included. Three tumoral regions were manually segmented: contrast-enhancing tumor, necrosis, and non-enhancing tumor. Radiomic features were extracted with a custom version of Pyradiomics and selected through Boruta algorithm. A Grid Search algorithm was applied when computing ten times K-fold cross-validation (K=10) to get the highest mean and lowest spread of accuracy. Model performance was assessed as AUC-ROC curve mean values with 95% confidence intervals (CI). Extreme Gradient Boosting (xGB) obtained highest accuracy for OS (74,5%), Adaboost (AB) for IDH mutation (87.5%), MGMT methylation (70,8%), Ki-67 expression (86%), and EGFR amplification (81%). Ensemble classifiers showed the best performance across tasks. High-scoring radiomic features shed light on possible correlations between MRI and tumor histology.

A Multiparametric MR-Based RadioFusionOmics Model with Robust Capabilities of Differentiating Glioblastoma Multiforme from Solitary Brain Metastasis.

Simple SummaryGlioblastoma multiforme (GBM) and solitary brain metastasis (SBM) are common brain tumors in adults. The two tumors often pose a diagnostic dilemma owing to their similar features on conventional magnetic resonance imaging (MRI). Ability to discriminate the two tumors is critical as it informs clinical treatment strategies. This pilot study attempts to employ the machine learning technique to identify GBM and SBM by fusing radiomics features of multiple MRI sequences and multiple models. A multiparametric MR-based RadioFusionOmics (RFO) model was developed and has demonstrated promising prediction accuracy for the identifications of GBM and SBM.This study aimed to evaluate the diagnostic potential of a novel RFO model in differentiating GBM and SBM with multiparametric MR sequences collected from 244 (131 GBM and 113 SBM) patients. Three basic volume of interests (VOIs) were delineated on the conventional axial MR images (T1WI, T2WI, T2_FLAIR, and CE_T1WI), including volumetric non-enhanced tumor (nET), enhanced tumor (ET), and peritumoral edema (pTE). Using the RFO model, radiomics features extracted from different multiparametric MRI sequence(s) and VOI(s) were fused and the best sequence and VOI, or possible combinations, were determined. A multi-disciplinary team (MDT)-like fusion was performed to integrate predictions from the high-performing models for the final discrimination of GBM vs. SBM. Image features extracted from the volumetric ET (VOIET) had dominant predictive performances over features from other VOI combinations. Fusion of VOIET features from the T1WI and T2_FLAIR sequences via the RFO model achieved a discrimination accuracy of AUC = 0.925, accuracy = 0.855, sensitivity = 0.856, and specificity = 0.853, on the independent testing cohort 1, and AUC = 0.859, accuracy = 0.836, sensitivity = 0.708, and specificity = 0.919 on the independent testing cohort 2, which significantly outperformed three experienced radiologists (p = 0.03, 0.01, 0.02, and 0.01, and p = 0.02, 0.01, 0.45, and 0.02, respectively) and the MDT-decision result of three experienced experts (p = 0.03, 0.02, 0.03, and 0.02, and p = 0.03, 0.02, 0.44, and 0.03, respectively).

Application of diffusion kurtosis imaging to the study of edema in solid and peritumoral areas of glioma

ObjectiveWhen gliomas grow in an infiltrative form, high-grade malignant glioma tissue extends beyond the contrast-enhancing tumor boundary, and this diffuse non-enhancing tumor infiltration is not visible on conventional MRI. The purpose of this study was to evaluate the of diffusion kurtosis imaging (DKI)-derived parameters in a group of patients with pre-operative gliomas, evaluating changes in the solid tumor and peritumoral edema area, and investigating their use for evaluating the recurrence and prognosis of gliomas. MethodsIn this retrospective study, 51 patients with gliomas who underwent biopsy or surgery underwent DKI scans before surgery. DKI scans were performed to generate DKI parameter maps of the solid tumor and peritumoral edema areas. In the solid tumor area, the kurtosis parameters showed the highest area under the curve (AUC), sensitivity, and specificity for distinguishing high- and low-grade gliomas (all P < 0.01). ResultsIn the peritumoral edema area, significant differences were found between groups with grade III and IV gliomas (P < 0.05). DKI parameters were found to correlate with clinical Ki-67 scores within the solid tumor area (MK: R2 = 0.288, P < 0.001; Kr: R2 = 0.270, P < 0.001; Ka: R2 = 0.274, P < 0.001; MD: R2 = 0.223, P < 0.001; FA: R2 = 0.098, P < 0.01). No significant correlations were found between Ki-67 and kurtosis parameters of peritumoral edema. ConclusionsIn this study, DKI showed potential utility for studying solid tumor and peritumoral edema of high grade gliomas.

CTNI-18. PHASE I AND PRELIMINARY PHASE 0 RESULTS OF ABTC 1801: A MULTI-ARM CLINICAL TRIAL OF THE PARP INHIBITOR PAMIPARIB (BGB290) WITH VERY LOW DOSE METRONOMIC TEMOZOLOMIDE IN RECURRENT IDH MUTANT GLIOMAS

Abstract BACKGROUND Preclinical studies have demonstrated that IDH1-mutant (IDHmt) gliomas harbor a BRCAness phenotype with a defect in homologous recombination that confers PARP inhibitor sensitivity. Pamiparib (BeiGene BGB-290) is an effective PARP-trapping PARP inhibitor with demonstrated favorable brain penetration in animal models. METHODS ABTC 1801 is a study examining the safety, pharmacokinetics, and efficacy of the combination of pamiparib with low dose metronomic temozolomide in recurrent IDHmt gliomas. The Phase I component utilized a 3 + 3 design with a target DLT rate ≤ 33%. Pamiparib dose was 60 mg BID and temozolomide dose 20 mg daily, with dose de-escalation levels for anticipated hematological toxicity. RESULTS Seven patients were enrolled on the Phase I portion at dose level 1; one patient was replaced for inadequate dosing secondary to non-compliance. All patients had prior radiotherapy and temozolomide; 4/7 had received multiple lines of alkylator therapy including nitrosoureas. Median age was 45, KPS 90, and number of prior relapses 3. Four patients had anaplastic astrocytoma, 2 anaplastic oligodendroglioma, and 1 glioblastoma. One of 6 patients (16.7%) experienced DLT during the first cycle (grade 3 neutropenia and thrombocytopenia). Two additional patients had grade 2 neutropenia. Two patients remain on study treatment at 12+ and 10+ months, while a third progressed at 10.1 months (PFS-6 43%). Tumor tissue was collected from two patients in the surgical arm. In enhancing and non-enhancing tumors, the mean unbound pamiparib concentrations were 198 and 160 nmol/L (or nmol/kg), respectively, which were &amp;gt; 20-fold the in vitro IC50 for PARP inhibition; mean unbound tumor-to-plasma ratios were 0.65 and 0.38. CONCLUSIONS Phase I results support pamiparib 60 mg BID with temozolomide 20 mg daily as the dosages for the Phase II study. Preliminary Phase 0 data suggest that pamiparib likely achieves sufficient pharmacologically active concentrations in both enhancing and non-enhancing brain tumors.

ITVT-08. Combined use of awake surgery and intraoperative MRI augments extent of resection in eloquent gliomas without neurological impairment

Abstract Maximizing the extent of resection (EOR) while preserving functional integrity is a mainstay of glioma surgery. Intraoperative MRI (iMRI) helps to augment the EOR. However, in eloquently located gliomas the significance of iMRI is controversial since the EOR is limited by functional rather than image-based boundaries. Thus, we sought to determine the impact of iMRI in our institutional series of awake glioma resections within or adjacent to eloquent (language, motor, sensory) areas since the implementation of a 1.5 Tesla iMRI in 2009. Tumor- and procedure-related data and functional outcome were assessed through medical charts review. The EOR was determined volumetrically on pre-, intra- and postoperative T1 contrast-enhanced (CE) and FLAIR MR images. 131 of 166 awake surgeries (79%) were performed under iMRI-guidance with concurrent language (n=72) and/or motor (n=50) mapping. iMRI was done when functional boundaries were reached (62%), for resection control (28%) or for other reasons (10%). Additional resection after iMRI (AR) was performed in 63 cases (73%); otherwise resection was terminated because the targeted EOR or functional boundaries were reached. New or deteriorated neurological deficits occurred in 20 patients prior and 15 patients post iMRI; however, all but 3 resolved within 6 months. Median EOR significantly increased after AR from 92.6% to 98.4% (∆5.8%; p&amp;lt;0.0001) in CE tumors and from 64.5% to 85.8% (∆21.3%; p&amp;lt;0.0001) in non-enhancing tumors. Remarkably, the reason to perform iMRI (resection control or functional limitations), did not affect the frequency of AR, deficits acquired post iMRI or the increase in EOR after AR. In conclusion, iMRI is a valuable adjunct to maximize the EOR in awake glioma resections without increasing the risk for functional impairment, particularly in non-enhancing tumors. Importantly, iMRI contributes to a maximized EOR even in cases where the resection had to be stopped because functional boundaries were reached.