MRI characteristics of brain gliomas have been used to predict clinical outcome and molecular tumor characteristics. However, previously reported imaging biomarkers have not been sufficiently accurate or reproducible to enter routine clinical practice and often rely on relatively simple MRI measures. The current study leverages advanced image analysis and machine learning algorithms to identify complex and reproducible imaging patterns predictive of overall survival and molecular subtype in glioblastoma (GB). One hundred five patients with GB were first used to extract approximately 60 diverse features from preoperative multiparametric MRIs. These imaging features were used by a machine learning algorithm to derive imaging predictors of patient survival and molecular subtype. Cross-validation ensured generalizability of these predictors to new patients. Subsequently, the predictors were evaluated in a prospective cohort of 29 new patients. Survival curves yielded a hazard ratio of 10.64 for predicted long versus short survivors. The overall, 3-way (long/medium/short survival) accuracy in the prospective cohort approached 80%. Classification of patients into the 4 molecular subtypes of GB achieved 76% accuracy. By employing machine learning techniques, we were able to demonstrate that imaging patterns are highly predictive of patient survival. Additionally, we found that GB subtypes have distinctive imaging phenotypes. These results reveal that when imaging markers related to infiltration, cell density, microvascularity, and blood–brain barrier compromise are integrated via advanced pattern analysis methods, they form very accurate predictive biomarkers. These predictive markers used solely preoperative images, hence they can significantly augment diagnosis and treatment of GB patients.
Abstract PURPOSE Multi-parametric MRI based radiomic signatures have highlighted the promise of artificial intelligence (AI) in neuro-oncology. However, inter-institution heterogeneity hinders generalization to data from unseen clinical institutions. To this end, we formulated the ReSPOND (Radiomics Signatures for PrecisiON Diagnostics) consortium for glioblastoma. Here, we seek non-invasive generalizable radiomic signatures from routine clinically-acquired MRI for prognostic stratification of glioblastoma patients. METHODS We identified a retrospective cohort of 606 patients with near/gross total tumor resection ( >90%), from 13 geographically-diverse institutions. All pre-operative structural MRI scans (T1,T1-Gd,T2,T2-FLAIR) were aligned to a common anatomical atlas. An automatic algorithm segmented the whole tumors (WTs) into 3 sub-compartments, i.e., enhancing (ET), necrotic core (NC), and peritumoral T2-FLAIR signal abnormality (ED). The combination of ET+NC defines the tumor core (TC). Quantitative radiomic features were extracted to generate our AI model to stratify patients into short- (< 14mts) and long-survivors ( >14mts). The model trained on 276 patients from a single institution was independently validated on 330 unseen patients from 12 left-out institutions, using the area-under-the-receiver-operating-characteristic-curve (AUC). RESULTS Each feature individually offered certain (limited but reproducible) value for identifying short-survivors: 1) TC closer to lateral ventricles (AUC=0.62); 2) larger ET/brain (AUC=0.61); 3) larger TC/brain (AUC=0.59); 4) larger WT/brain (AUC=0.55); 5) larger ET/WT (AUC=0.59); 6) smaller ED/WT (AUC=0.57); 7) larger ventricle deformations (AUC=0.6). Integrating all features and age, through a multivariate AI model, resulted in higher accuracy (AUC=0.7; 95% C.I.,0.64-0.77). CONCLUSION Prognostic stratification using basic radiomic features is highly reproducible across diverse institutions and patient populations. Multivariate integration yields relatively more accurate and generalizable radiomic signatures, across institutions. Our results offer promise for generalizable non-invasive in vivo signatures of survival prediction in patients with glioblastoma. Extracted features from clinically-acquired imaging, renders these signatures easier for clinical translation. Large-scale evaluation could contribute to improving patient management and treatment planning. *Indicates equal authorship.
We describe the design and results from the BraTS 2023 Intracranial Meningioma Segmentation Challenge. The BraTS Meningioma Challenge differed from prior BraTS Glioma challenges in that it focused on meningiomas, which are typically benign extra-axial tumors with diverse radiologic and anatomical presentation and a propensity for multiplicity. Nine participating teams each developed deep-learning automated segmentation models using image data from the largest multi-institutional systematically expert annotated multilabel multi-sequence meningioma MRI dataset to date, which included 1000 training set cases, 141 validation set cases, and 283 hidden test set cases. Each case included T2, T2/FLAIR, T1, and T1Gd brain MRI sequences with associated tumor compartment labels delineating enhancing tumor, non-enhancing tumor, and surrounding non-enhancing T2/FLAIR hyperintensity. Participant automated segmentation models were evaluated and ranked based on a scoring system evaluating lesion-wise metrics including dice similarity coefficient (DSC) and 95% Hausdorff Distance. The top ranked team had a lesion-wise median dice similarity coefficient (DSC) of 0.976, 0.976, and 0.964 for enhancing tumor, tumor core, and whole tumor, respectively and a corresponding average DSC of 0.899, 0.904, and 0.871, respectively. These results serve as state-of-the-art benchmarks for future pre-operative meningioma automated segmentation algorithms. Additionally, we found that 1286 of 1424 cases (90.3%) had at least 1 compartment voxel abutting the edge of the skull-stripped image edge, which requires further investigation into optimal pre-processing face anonymization steps.
Imaging and autopsy studies show intracranial gadolinium deposition in patients who have undergone serial contrast-enhanced MRIs. This observation has raised concerns when using contrast administration in patients who receive frequent MRIs. To address this, we implemented a contrast-conditional protocol wherein gadolinium is administered only for multiple sclerosis (MS) patients with imaging evidence of new disease activity on precontrast imaging. In this study, we explore the economic impact of our new MRI protocol.We compared scanner time and Medicare reimbursement using our contrast-conditional methodology versus that of prior protocols where all patients received gadolinium.For 422 patients over 4 months, the contrast-conditional protocol amounted to 60% decrease in contrast injection and savings of approximately 20% of MRI scanner time. If the extra scanner time was used for performing MS follow-up MRIs in additional patients, the contrast-conditional protocol would amount to net revenue loss of $21,707 (∼3.7%).Implementation of a new protocol to limit contrast in MS follow-up MRIs led to a minimal decrease in revenue when controlled for scanner time utilized and is outweighed by other benefits, including substantial decreased gadolinium administration, increased patient comfort, and increased availability of scanner time, which depending on type of studies performed could result in additional financial benefit.
The detection and monitoring of brain lesions caused by multiple sclerosis is commonly performed with the use of magnetic resonance imaging. Analysis of a large number of images is a time-consuming challenge to the neuroradiologist, that can be accelerated with the assistance of computer-detection software. In 98 baseline and follow-up brain magnetic resonance studies from 88 patients with a diagnosis of multiple sclerosis, we employed locally developed lesion-detection software to assess temporal change in the load of brain lesions and compared its results to routine clinical reports. Analyzing the differences between the follow-up study and the baseline study, the software displays the results in the form of a scrollable axial volume, with the changed lesions highlighted in different colors and superimposed on the baseline reference scan. Disagreements between the software and the clinical readers in the detection of changed lesions were observed only in 11 (11.2%) cases, and the difference did not reach statistical significance (p=0.07). The mean interpretation time with assistance of the software was 2.7±2.2 minutes. We conclude that the performance of the software-assisted interpretation in the analysis of change over time in multiple sclerosis brain lesions is comparable to the performance of clinical readers, with a possibly shorter assessment time. Our study demonstrates the potential of including lesion-detection software in the workflow of neuroradiology practice.
Glioblastoma, the most common and aggressive adult brain tumor, is considered non-curative at diagnosis. Current literature shows promise on imaging-based overall survival prediction for patients with glioblastoma while integrating advanced (structural, perfusion, and diffusion) multiparametric magnetic resonance imaging (AdvmpMRI). However, most patients prior to initiation of therapy typically undergo only basic structural mpMRI (BasmpMRI, i.e., T1,T1-Gd,T2,T2-FLAIR) pre-operatively, rather than Adv-mpMRI. Here we assess a retrospective cohort of 101 glioblastoma patients with available Adv-mpMRI from a previous study, which has shown that an initial feature panel (IFP) extracted from Adv-mpMRI can yield accurate overall survival stratification. We further focus on demonstrating that equally accurate prediction models can be constructed using augmented feature panels (AFP) extracted solely from Bas-mpMRI, obviating the need for using Adv-mpMRI. The classification accuracy of the model utilizing Adv-mpMRI protocols and the IFP was 72.77%, and improved to 74.26% when utilizing the AFP on Bas-mpMRI. Furthermore, Kaplan-Meier analysis demonstrated superior classification of subjects into short-, intermediate-, and long-survivor classes when using AFP on Basic-mpMRI. This quantitative evaluation indicates that accurate survival prediction in glioblastoma patients is feasible by using solely Bas-mpMRI and integrative radiomic analysis can compensate for the lack of Adv-mpMRI. Our finding holds promise for predicting overall survival based on commonly-acquired Bas-mpMRI, and hence for potential generalization across multiple institutions that may not have access to Adv-mpMRI, facilitating better patient selection.
The aim of this study was to explore whether intellectual performance in children with Sickle Cell Disease and with low risk of stroke as determined with conventional transcranial Doppler ultrasonography (TCD) criteria was associated with hemodynamic parameters in imaging TCD, when controlling for hematological and socio-economical variables and presence of silent infarcts. We performed neuropsychological testing with Kaufman Brief Intelligence Test (K-BIT-IQ) and imaging TCD examinations to measure blood flow velocities and pulsatility indexes (PI) in the middle cerebral arteries (MCA) In 46 children with homozygous HbSS (mean age 108±34 months, range limits: 47–166 months; 24 females), without a history of stroke or transient ischemic attack, with no stenosis on magnetic resonance angiography and with velocities below 170 cm/s in screening conventional TCD. Mean K-BIT IQ Composite and Vocabulary scores (91±13 and 86±14 respectively) were significantly below the average scores of 100 for the age-matched population (one sample t-test=5.21, p<0.001). Using univariate and multivariate regression models, we found that lower PI in the right MCA was associated with lower K-BIT-IQ Composite and Vocabulary scores. Furthermore, we found that interhemispheric differences in PIs were even more strongly associated with neuropsychological performance, whereas flow velocities were not associated with the K-BIT-IQ score. Using a model of chronic anemia, we found that cognitive functioning was associated with cerebral hemodynamics.
In treating glioblastoma (GB), surgical and chemotherapeutic treatment guidelines are, for the most part, independent of tumor location. In this work, we compiled imaging data from a large cohort of GB patients to create statistical atlases illustrating the disease spatial frequency as a function of patient demographics as well as tumor characteristics. Two-hundred-six patients with pathology-proven glioblastoma were included. Of those, 65 had pathology-proven recurrence and 113 had molecular subtype and genetic information. We used validated software to segment the tumors in all patients and map them from patient space into a common template. We then created statistical maps that described the spatial location of tumors with respect to demographics and tumor characteristics. We applied a chi-square test to determine whether pattern differences were statistically significant. The most frequent location for glioblastoma in our patient population is the right temporal lobe. There are statistically significant differences when comparing patterns using demographic data such as gender (p = 0.0006) and age (p = 0.006). Small and large tumors tend to occur in separate locations (p = 0.0007). The tumors tend to occur in different locations according to their molecular subtypes (p < 10− 6). The classical subtype tends to spare the frontal lobes, the neural subtype tend to involve the inferior right frontal lobe. Although the sample size is limited, there was a difference in location according to EGFR VIII genotype (p < 10− 4), with a right temporal dominance for EFGR VIII negative tumors, and frontal lobe dominance in EGFR VIII positive tumors. Spatial location of GB is an important factor that correlates with demographic factors and tumor characteristics, which should therefore be considered when evaluating a patient with GB and might assist in personalized treatment.
Although machine learning (ML) has shown promise in numerous domains, there are concerns about generalizability to out-of-sample data. This is currently addressed by centrally sharing ample, and importantly diverse, data from multiple sites. However, such centralization is challenging to scale (or even not feasible) due to various limitations. Federated ML (FL) provides an alternative to train accurate and generalizable ML models, by only sharing numerical model updates. Here we present findings from the largest FL study to-date, involving data from 71 healthcare institutions across 6 continents, to generate an automatic tumor boundary detector for the rare disease of glioblastoma, utilizing the largest dataset of such patients ever used in the literature (25,256 MRI scans from 6,314 patients). We demonstrate a 33% improvement over a publicly trained model to delineate the surgically targetable tumor, and 23% improvement over the tumor's entire extent. We anticipate our study to: 1) enable more studies in healthcare informed by large and diverse data, ensuring meaningful results for rare diseases and underrepresented populations, 2) facilitate further quantitative analyses for glioblastoma via performance optimization of our consensus model for eventual public release, and 3) demonstrate the effectiveness of FL at such scale and task complexity as a paradigm shift for multi-site collaborations, alleviating the need for data sharing.
Patients with multiple sclerosis routinely have MR imaging with contrast every 6-12 months to assess response to medication. Multiple recent studies provide evidence of tissue deposition of MR imaging contrast agents, questioning the long-term safety of these agents. The goal of this retrospective image-analysis study was to determine whether contrast could be reserved for only those patients who show new MS lesions on follow-up examinations.We retrospectively reviewed brain MRIs of 138 patients. To increase our sensitivity, we used a previously described computerized image-comparison software to evaluate the stability or progression of multiple sclerosis white matter lesions in noncontrast FLAIR sequences. We correlated these findings with evidence of contrast-enhancing lesions on the enhanced T1 sequence from the same scan.Thirty-three scans showed an increase in white matter lesion burden. Among those 33 patients, 14 examinations also demonstrated enhancing new lesions. While we found a single example of enhancement of a pre-existing white matter lesion that appeared unchanged in size, that same examination showed an overall increase in lesion burden with enhancement of other, new lesions. Thus, we found that all patients with enhancing lesions had evidence of progression on their noncontrast imaging.Because all enhancing lesions were associated with new lesions on unenhanced imaging and progression was only evident in 24% of patients, in patients with relapsing-remitting MS, it is reasonable to consider reserving contrast for only those patients with evidence of progression on noncontrast MR images.
