Segmentation of glioblastomas in early post-operative multi-modal MRI with deep neural networks.

Extent of resection after surgery is one of the main prognostic factors for patients diagnosed with glioblastoma. To achieve this, accurate segmentation and classification of residual tumor from post-operative MR images is essential. The current standard method for estimating it is subject to high inter- and intra-rater variability, and an automated method for segmentation of residual tumor in early post-operative MRI could lead to a more accurate estimation of extent of resection. In this study, two state-of-the-art neural network architectures for pre-operative segmentation were trained for the task. The models were extensively validated on a multicenter dataset with nearly 1000 patients, from 12 hospitals in Europe and the United States. The best performance achieved was a 61% Dice score, and the best classification performance was about 80% balanced accuracy, with a demonstrated ability to generalize across hospitals. In addition, the segmentation performance of the best models was on par with human expert raters. The predicted segmentations can be used to accurately classify the patients into those with residual tumor, and those with gross total resection.

Multi-class glioma segmentation on real-world data with missing MRI sequences: comparison of three deep learning algorithms.

This study tests the generalisability of three Brain Tumor Segmentation (BraTS) challenge models using a multi-center dataset of varying image quality and incomplete MRI datasets. In this retrospective study, DeepMedic, no-new-Unet (nn-Unet), and NVIDIA-net (nv-Net) were trained and tested using manual segmentations from preoperative MRI of glioblastoma (GBM) and low-grade gliomas (LGG) from the BraTS 2021 dataset (1251 in total), in addition to 275 GBM and 205 LGG acquired clinically across 12 hospitals worldwide. Data was split into 80% training, 5% validation, and 15% internal test data. An additional external test-set of 158 GBM and 69 LGG was used to assess generalisability to other hospitals' data. All models' median Dice similarity coefficient (DSC) for both test sets were within, or higher than, previously reported human inter-rater agreement (range of 0.74-0.85). For both test sets, nn-Unet achieved the highest DSC (internal = 0.86, external = 0.93) and the lowest Hausdorff distances (10.07, 13.87 mm, respectively) for all tumor classes (p < 0.001). By applying Sparsified training, missing MRI sequences did not statistically affect the performance. nn-Unet achieves accurate segmentations in clinical settings even in the presence of incomplete MRI datasets. This facilitates future clinical adoption of automated glioma segmentation, which could help inform treatment planning and glioma monitoring.

Combined molecular subtyping, grading, and segmentation of glioma using multi-task deep learning.

BACKGROUND: Accurate characterization of glioma is crucial for clinical decision making. A delineation of the tumor is also desirable in the initial decision stages but is time-consuming. Previously, deep learning methods have been developed that can either non-invasively predict the genetic or histological features of glioma, or that can automatically delineate the tumor, but not both tasks at the same time. Here, we present our method that can predict the molecular subtype and grade, while simultaneously providing a delineation of the tumor. METHODS: We developed a single multi-task convolutional neural network that uses the full 3D, structural, preoperative MRI scans to predict the IDH mutation status, the 1p/19q co-deletion status, and the grade of a tumor, while simultaneously segmenting the tumor. We trained our method using a patient cohort containing 1508 glioma patients from 16 institutes. We tested our method on an independent dataset of 240 patients from 13 different institutes. RESULTS: In the independent test set, we achieved an IDH-AUC of 0.90, an 1p/19q co-deletion AUC of 0.85, and a grade AUC of 0.81 (grade II/III/IV). For the tumor delineation, we achieved a mean whole tumor Dice score of 0.84. CONCLUSIONS: We developed a method that non-invasively predicts multiple, clinically relevant features of glioma. Evaluation in an independent dataset shows that the method achieves a high performance and that it generalizes well to the broader clinical population. This first-of-its-kind method opens the door to more generalizable, instead of hyper-specialized, AI methods.

Preoperative Brain Tumor Imaging: Models and Software for Segmentation and Standardized Reporting.

For patients suffering from brain tumor, prognosis estimation and treatment decisions are made by a multidisciplinary team based on a set of preoperative MR scans. Currently, the lack of standardized and automatic methods for tumor detection and generation of clinical reports, incorporating a wide range of tumor characteristics, represents a major hurdle. In this study, we investigate the most occurring brain tumor types: glioblastomas, lower grade gliomas, meningiomas, and metastases, through four cohorts of up to 4,000 patients. Tumor segmentation models were trained using the AGU-Net architecture with different preprocessing steps and protocols. Segmentation performances were assessed in-depth using a wide-range of voxel and patient-wise metrics covering volume, distance, and probabilistic aspects. Finally, two software solutions have been developed, enabling an easy use of the trained models and standardized generation of clinical reports: Raidionics and Raidionics-Slicer. Segmentation performances were quite homogeneous across the four different brain tumor types, with an average true positive Dice ranging between 80 and 90%, patient-wise recall between 88 and 98%, and patient-wise precision around 95%. In conjunction to Dice, the identified most relevant other metrics were the relative absolute volume difference, the variation of information, and the Hausdorff, Mahalanobis, and object average symmetric surface distances. With our Raidionics software, running on a desktop computer with CPU support, tumor segmentation can be performed in 16-54 s depending on the dimensions of the MRI volume. For the generation of a standardized clinical report, including the tumor segmentation and features computation, 5-15 min are necessary. All trained models have been made open-access together with the source code for both software solutions and validation metrics computation. In the future, a method to convert results from a set of metrics into a final single score would be highly desirable for easier ranking across trained models. In addition, an automatic classification of the brain tumor type would be necessary to replace manual user input. Finally, the inclusion of post-operative segmentation in both software solutions will be key for generating complete post-operative standardized clinical reports.

Predictors for time to awake in patients undergoing awake craniotomies.

OBJECTIVE: Awake craniotomies are often characterized by alternating asleep-awake-asleep periods. Preceding the awake phase, patients are weaned from anesthesia and mechanical ventilation. Although clinicians aim to minimize the time to awake for patient safety and operating room efficiency, in some patients, the time to awake exceeds 20 minutes. The goal of this study was to determine the average time to awake and the factors associated with prolonged time to awake (> 20 minutes) in patients undergoing awake craniotomy. METHODS: Records of patients who underwent awake craniotomy between 2003 and 2020 were evaluated. Time to awake was defined as the time between discontinuation of propofol and remifentanil infusion and the time of extubation. Patient and perioperative characteristics were explored as predictors for time to awake using logistic regression analyses. RESULTS: Data of 307 patients were analyzed. The median (IQR) time to awake was 13 (10-20) minutes and exceeded 20 minutes in 17% (95% CI 13%-21%) of the patients. In both univariate and multivariable analyses, increased age, nonsmoker status, and American Society of Anesthesiologists (ASA) class III versus II were associated with a time to awake exceeding 20 minutes. BMI, as well as the use of alcohol, drugs, dexamethasone, or antiepileptic agents, was not significantly associated with the time to awake. CONCLUSIONS: While most patients undergoing awake craniotomy are awake within a reasonable time frame after discontinuation of propofol and remifentanil infusion, time to awake exceeded 20 minutes in 17% of the patients. Increasing age, nonsmoker status, and higher ASA classification were found to be associated with a prolonged time to awake.

On the cutting edge of glioblastoma surgery: where neurosurgeons agree and disagree on surgical decisions.

OBJECTIVE: The aim of glioblastoma surgery is to maximize the extent of resection while preserving functional integrity. Standards are lacking for surgical decision-making, and previous studies indicate treatment variations. These shortcomings reflect the need to evaluate larger populations from different care teams. In this study, the authors used probability maps to quantify and compare surgical decision-making throughout the brain by 12 neurosurgical teams for patients with glioblastoma. METHODS: The study included all adult patients who underwent first-time glioblastoma surgery in 2012-2013 and were treated by 1 of the 12 participating neurosurgical teams. Voxel-wise probability maps of tumor location, biopsy, and resection were constructed for each team to identify and compare patient treatment variations. Brain regions with different biopsy and resection results between teams were identified and analyzed for patient functional outcome and survival. RESULTS: The study cohort consisted of 1087 patients, of whom 363 underwent a biopsy and 724 a resection. Biopsy and resection decisions were generally comparable between teams, providing benchmarks for probability maps of resections and biopsies for glioblastoma. Differences in biopsy rates were identified for the right superior frontal gyrus and indicated variation in biopsy decisions. Differences in resection rates were identified for the left superior parietal lobule, indicating variations in resection decisions. CONCLUSIONS: Probability maps of glioblastoma surgery enabled capture of clinical practice decisions and indicated that teams generally agreed on which region to biopsy or to resect. However, treatment variations reflecting clinical dilemmas were observed and pinpointed by using the probability maps, which could therefore be useful for quality-of-care discussions between surgical teams for patients with glioblastoma.

Glioblastoma Surgery Imaging-Reporting and Data System: Validation and Performance of the Automated Segmentation Task.

For patients with presumed glioblastoma, essential tumor characteristics are determined from preoperative MR images to optimize the treatment strategy. This procedure is time-consuming and subjective, if performed by crude eyeballing or manually. The standardized GSI-RADS aims to provide neurosurgeons with automatic tumor segmentations to extract tumor features rapidly and objectively. In this study, we improved automatic tumor segmentation and compared the agreement with manual raters, describe the technical details of the different components of GSI-RADS, and determined their speed. Two recent neural network architectures were considered for the segmentation task: nnU-Net and AGU-Net. Two preprocessing schemes were introduced to investigate the tradeoff between performance and processing speed. A summarized description of the tumor feature extraction and standardized reporting process is included. The trained architectures for automatic segmentation and the code for computing the standardized report are distributed as open-source and as open-access software. Validation studies were performed on a dataset of 1594 gadolinium-enhanced T1-weighted MRI volumes from 13 hospitals and 293 T1-weighted MRI volumes from the BraTS challenge. The glioblastoma tumor core segmentation reached a Dice score slightly below 90%, a patientwise F1-score close to 99%, and a 95th percentile Hausdorff distance slightly below 4.0 mm on average with either architecture and the heavy preprocessing scheme. A patient MRI volume can be segmented in less than one minute, and a standardized report can be generated in up to five minutes. The proposed GSI-RADS software showed robust performance on a large collection of MRI volumes from various hospitals and generated results within a reasonable runtime.

Glioblastoma Surgery Imaging-Reporting and Data System: Standardized Reporting of Tumor Volume, Location, and Resectability Based on Automated Segmentations.

Treatment decisions for patients with presumed glioblastoma are based on tumor characteristics available from a preoperative MR scan. Tumor characteristics, including volume, location, and resectability, are often estimated or manually delineated. This process is time consuming and subjective. Hence, comparison across cohorts, trials, or registries are subject to assessment bias. In this study, we propose a standardized Glioblastoma Surgery Imaging Reporting and Data System (GSI-RADS) based on an automated method of tumor segmentation that provides standard reports on tumor features that are potentially relevant for glioblastoma surgery. As clinical validation, we determine the agreement in extracted tumor features between the automated method and the current standard of manual segmentations from routine clinical MR scans before treatment. In an observational consecutive cohort of 1596 adult patients with a first time surgery of a glioblastoma from 13 institutions, we segmented gadolinium-enhanced tumor parts both by a human rater and by an automated algorithm. Tumor features were extracted from segmentations of both methods and compared to assess differences, concordance, and equivalence. The laterality, contralateral infiltration, and the laterality indices were in excellent agreement. The native and normalized tumor volumes had excellent agreement, consistency, and equivalence. Multifocality, but not the number of foci, had good agreement and equivalence. The location profiles of cortical and subcortical structures were in excellent agreement. The expected residual tumor volumes and resectability indices had excellent agreement, consistency, and equivalence. Tumor probability maps were in good agreement. In conclusion, automated segmentations are in excellent agreement with manual segmentations and practically equivalent regarding tumor features that are potentially relevant for neurosurgical purposes. Standard GSI-RADS reports can be generated by open access software.

Timing of glioblastoma surgery and patient outcomes: a multicenter cohort study.

BACKGROUND: The impact of time-to-surgery on clinical outcome for patients with glioblastoma has not been determined. Any delay in treatment is perceived as detrimental, but guidelines do not specify acceptable timings. In this study, we relate the time to glioblastoma surgery with the extent of resection and residual tumor volume, performance change, and survival, and we explore the identification of patients for urgent surgery. METHODS: Adults with first-time surgery in 2012-2013 treated by 12 neuro-oncological teams were included in this study. We defined time-to-surgery as the number of days between the diagnostic MR scan and surgery. The relation between time-to-surgery and patient and tumor characteristics was explored in time-to-event analysis and proportional hazard models. Outcome according to time-to-surgery was analyzed by volumetric measurements, changes in performance status, and survival analysis with patient and tumor characteristics as modifiers. RESULTS: Included were 1033 patients of whom 729 had a resection and 304 a biopsy. The overall median time-to-surgery was 13 days. Surgery was within 3 days for 235 (23%) patients, and within a month for 889 (86%). The median volumetric doubling time was 22 days. Lower performance status (hazard ratio [HR] 0.942, 95% confidence interval [CI] 0.893-0.994) and larger tumor volume (HR 1.012, 95% CI 1.010-1.014) were independently associated with a shorter time-to-surgery. Extent of resection, residual tumor volume, postoperative performance change, and overall survival were not associated with time-to-surgery. CONCLUSIONS: With current decision-making for urgent surgery in selected patients with glioblastoma and surgery typically within 1 month, we found equal extent of resection, residual tumor volume, performance status, and survival after longer times-to-surgery.

Accurate MR Image Registration to Anatomical Reference Space for Diffuse Glioma.

To summarize the distribution of glioma location within a patient population, registration of individual MR images to anatomical reference space is required. In this study, we quantified the accuracy of MR image registration to anatomical reference space with linear and non-linear transformations using estimated tumor targets of glioblastoma and lower-grade glioma, and anatomical landmarks at pre- and post-operative time-points using six commonly used registration packages (FSL, SPM5, DARTEL, ANTs, Elastix, and NiftyReg). Routine clinical pre- and post-operative, post-contrast T1-weighted images of 20 patients with glioblastoma and 20 with lower-grade glioma were collected. The 2009a Montreal Neurological Institute brain template was used as anatomical reference space. Tumors were manually segmented in the patient space and corresponding healthy tissue was delineated as a target volume in the anatomical reference space. Accuracy of the tumor alignment was quantified using the Dice score and the Hausdorff distance. To measure the accuracy of general brain alignment, anatomical landmarks were placed in patient and in anatomical reference space, and the landmark distance after registration was quantified. Lower-grade gliomas were registered more accurately than glioblastoma. Registration accuracy for pre- and post-operative MR images did not differ. SPM5 and DARTEL registered tumors most accurate, and FSL least accurate. Non-linear transformations resulted in more accurate general brain alignment than linear transformations, but tumor alignment was similar between linear and non-linear transformation. We conclude that linear transformation suffices to summarize glioma locations in anatomical reference space.

Quantifying eloquent locations for glioblastoma surgery using resection probability maps.

OBJECTIVE: Decisions in glioblastoma surgery are often guided by presumed eloquence of the tumor location. The authors introduce the "expected residual tumor volume" (eRV) and the "expected resectability index" (eRI) based on previous decisions aggregated in resection probability maps. The diagnostic accuracy of eRV and eRI to predict biopsy decisions, resectability, functional outcome, and survival was determined. METHODS: Consecutive patients with first-time glioblastoma surgery in 2012-2013 were included from 12 hospitals. The eRV was calculated from the preoperative MR images of each patient using a resection probability map, and the eRI was derived from the tumor volume. As reference, Sawaya's tumor location eloquence grades (EGs) were classified. Resectability was measured as observed extent of resection (EOR) and residual volume, and functional outcome as change in Karnofsky Performance Scale score. Receiver operating characteristic curves and multivariable logistic regression were applied. RESULTS: Of 915 patients, 674 (74%) underwent a resection with a median EOR of 97%, functional improvement in 71 (8%), functional decline in 78 (9%), and median survival of 12.8 months. The eRI and eRV identified biopsies and EORs of at least 80%, 90%, or 98% better than EG. The eRV and eRI predicted observed residual volumes under 10, 5, and 1 ml better than EG. The eRV, eRI, and EG had low diagnostic accuracy for functional outcome changes. Higher eRV and lower eRI were strongly associated with shorter survival, independent of known prognostic factors. CONCLUSIONS: The eRV and eRI predict biopsy decisions, resectability, and survival better than eloquence grading and may be useful preoperative indices to support surgical decisions.

Robust Deep Learning-based Segmentation of Glioblastoma on Routine Clinical MRI Scans Using Sparsified Training.

PURPOSE: To improve the robustness of deep learning-based glioblastoma segmentation in a clinical setting with sparsified datasets. MATERIALS AND METHODS: In this retrospective study, preoperative T1-weighted, T2-weighted, T2-weighted fluid-attenuated inversion recovery, and postcontrast T1-weighted MRI from 117 patients (median age, 64 years; interquartile range [IQR], 55-73 years; 76 men) included within the Multimodal Brain Tumor Image Segmentation (BraTS) dataset plus a clinical dataset (2012-2013) with similar imaging modalities of 634 patients (median age, 59 years; IQR, 49-69 years; 382 men) with glioblastoma from six hospitals were used. Expert tumor delineations on the postcontrast images were available, but for various clinical datasets, one or more sequences were missing. The convolutional neural network, DeepMedic, was trained on combinations of complete and incomplete data with and without site-specific data. Sparsified training was introduced, which randomly simulated missing sequences during training. The effects of sparsified training and center-specific training were tested using Wilcoxon signed rank tests for paired measurements. RESULTS: A model trained exclusively on BraTS data reached a median Dice score of 0.81 for segmentation on BraTS test data but only 0.49 on the clinical data. Sparsified training improved performance (adjusted P < .05), even when excluding test data with missing sequences, to median Dice score of 0.67. Inclusion of site-specific data during sparsified training led to higher model performance Dice scores greater than 0.8, on par with a model based on all complete and incomplete data. For the model using BraTS and clinical training data, inclusion of site-specific data or sparsified training was of no consequence. CONCLUSION: Accurate and automatic segmentation of glioblastoma on clinical scans is feasible using a model based on large, heterogeneous, and partially incomplete datasets. Sparsified training may boost the performance of a smaller model based on public and site-specific data.Supplemental material is available for this article.Published under a CC BY 4.0 license.

Comparing Glioblastoma Surgery Decisions Between Teams Using Brain Maps of Tumor Locations, Biopsies, and Resections.

PURPOSE: The aim of glioblastoma surgery is to maximize the extent of resection while preserving functional integrity, which depends on the location within the brain. A standard to compare these decisions is lacking. We present a volumetric voxel-wise method for direct comparison between two multidisciplinary teams of glioblastoma surgery decisions throughout the brain. METHODS: Adults undergoing first-time glioblastoma surgery from 2012 to 2013 performed by two neuro-oncologic teams were included. Patients had had a diagnostic biopsy or resection. Preoperative tumors and postoperative residues were segmented on magnetic resonance imaging in three dimensions and registered to standard brain space. Voxel-wise probability maps of tumor location, biopsy, and resection were constructed for each team to compare patient referral bias, indication variation, and treatment variation. To evaluate the quality of care, subgroups of differentially resected brain regions were analyzed for survival and functional outcome. RESULTS: One team included 101 patients, and the other included 174; 91 tumors were biopsied, and 181 were resected. Patient characteristics were largely comparable between teams. Distributions of tumor locations were dissimilar, suggesting referral bias. Distributions of biopsies were similar, suggesting absence of indication variation. Differentially resected regions were identified in the anterior limb of the right internal capsule and the right caudate nucleus, indicating treatment variation. Patients with (n = 12) and without (n = 6) surgical removal in these regions had similar overall survival and similar permanent neurologic deficits. CONCLUSION: Probability maps of tumor location, biopsy, and resection provide additional information that can inform surgical decision making across multidisciplinary teams for patients with glioblastoma.

Earliest radiological progression in glioblastoma by multidisciplinary consensus review.

BACKGROUND: Detection of glioblastoma progression is important for clinical decision-making on cessation or initiation of therapy, for enrollment in clinical trials, and for response measurement in time and location. The RANO-criteria are considered standard for the timing of progression. To evaluate local treatment, we aim to find the most accurate progression location. We determined the differences in progression free survival (PFS) and in tumor volumes at progression (Vprog) by three definitions of progression. METHODS: In a consecutive cohort of 73 patients with newly-diagnosed glioblastoma between 1/1/2012 and 31/12/2013, progression was established according to three definitions. We determined (1) earliest radiological progression (ERP) by retrospective multidisciplinary consensus review using all available imaging and follow-up, (2) clinical practice progression (CPP) from multidisciplinary tumor board conclusions, and (3) progression by the RANO-criteria. RESULTS: ERP was established in 63 (86%), CPP in 64 (88%), RANO progression in 42 (58%). Of the 63 patients who had died, 37 (59%) did with prior RANO-progression, compared to 57 (90%) for both ERP and CPP. The median overall survival was 15.3 months. The median PFS was 8.8 months for ERP, 9.5 months for CPP, and 11.8 months for RANO. The PFS by ERP was shorter than CPP (HR 0.57, 95% CI 0.38-0.84, p = 0.004) and RANO-progression (HR 0.29, 95% CI 0.19-0.43, p < 0.001). The Vprog were significantly smaller for ERP (median 8.8 mL), than for CPP (17 mL) and RANO (22 mL). CONCLUSION: PFS and Vprog vary considerably between progression definitions. Earliest radiological progression by retrospective consensus review should be considered to accurately localize progression and to address confounding of lead time bias in clinical trial enrollment.