Prostate-specific membrane antigen (PSMA) is expressed in several solid tumours, including brain metastases (BMs) from lung cancer. We investigated quantitative [18F]PSMA-1007 uptake and treatment response in BMs of patients with non-small cell lung cancer (NSCLC). Eleven patients with BM of NSCLC underwent 95-min dual-time-point dynamic [18F]PSMA-1007 PET/CT and a whole-body static scan before and after stereotactic radiosurgery (SRS) with a median dose of 20 Gy (range: 8–25 Gy) given in a median of 1 fraction (range: 1–3). Seven patients completed both exams. Image-derived input functions were extracted from the internal carotid arteries, and pharmacokinetic analysis using Patlak modelling yielded the influx constant (Ki). Standardised uptake value (SUV), biological tumour volume (BTV) and tumour heterogeneity using the coefficient of variance (CoV) were assessed. [18F]PSMA-1007 PET showed uptake in the first exam in BMs and primary lung tumour in all patients. Significant inter-patient and intra-lesion heterogeneity in tracer uptake was observed in BMs with median Ki of 0.005 ml/ccm/min (range: 0.003–0.018 ml/ccm/min), SUVpeak of 4.2 g/ml (range: 0.4–34.4 g/ml), CoV of 0.36 (range 0.06–0.85) and BTV of 2.35 ml (range 0.03–22.29 ml). After SRS, reductions in Ki (37 https://clinicaltrials.gov/ct2/show/NCT03951142 . EudraCT no 2018-003229-27. Registered 26 February 2019, https://www.clinicaltrialsregister.eu/ctr-search/trial/2018-003229-27/NO .

Rigorous dataset partitioning is a foundational, yet frequently overlooked, prerequisite for reliable deep learning in longitudinal medical imaging. Naively shuffling small clinical cohorts routinely introduces covariate shifts and temporal sampling imbalances across training, validation, and test subsets, exposing downstream models to out-of-distribution evaluation. We address this vulnerability with an auditable Tripartite Dataset Analytics Framework that systematically characterizes spatial grid integrity, multi-parametric intensity fingerprints, and longitudinal temporal trajectories, quantifying the heavy-tailed feature dispersion and irregular, episodic sampling intervals typical of real-world clinical cohorts. Building on this characterization, we formalize an unsupervised spatio-temporal cohort-balancing standard operating procedure (SOP) that combines elbow-optimized K-means clustering over a standardized, six-dimensional joint intensity-temporal feature space with intra-cluster proportionate stratified sampling. On a longitudinal, contrast-enhanced T1-weighted brain MRI cohort (N=149), the protocol reduces the maximum cross-subset intensity bias from 34.1

Instance-level lesion detection has been an increasingly larger focal point in medical image segmentation besides the more standard voxel-level overlap. Still, most pipelines are trained and post-processed for voxel overlap alone. In particular, the mismatch is most pronounced for small lesions, where a near-miss prediction—substantial overlap that falls just short of the instance-matching threshold—scores the same as a complete miss. In our ISLES'26 submission, we found that closing this gap mattered far more in post-processing than in architecture design. Our Volume-Conditioned Adaptive Post-Processing (VCAP) scheme adjusts component-size thresholds to each case's predicted lesion burden, improving Lesion-F1 by 0.032 (unbiased cross-fold estimate)—approximately 6 times larger than any architectural change we tested. A resolution-aware attention architecture (Viola2Plus), designed for small-lesion segmentation, shows why the distinction matters: it left small-lesion Dice unchanged but raised small-lesion detection rate by 3.7%, a real effect voxel-overlap metrics alone would have missed. Under 5-fold cross-validation on the 1,453-case training set, our post-processed two-architecture ensemble achieves Dice 0.651 and Lesion-F1 0.614, versus 0.644 and 0.573 for the unprocessed single-model baseline.

Abstract Molecular intratumoral heterogeneity is a defining feature of adult diffuse gliomas, yet how it relates to tissue biomechanics remains incompletely understood. Our prior magnetic resonance elastography (MRE) studies showed that regions of increased stiffness in glioblastoma are linked to extracellular matrix (ECM) remodeling and adverse prognosis, indicating that MRE captures key aspects of tumor physiology with prognostic value. Here, we extend this framework by integrating MRE with multi-omic profiling of multiregional tumor biopsies from adult diffuse gliomas to further characterize tissue mechanics and their relation to glioma biology. Using MRE, we show that local glioma mechanics can be summarized into three states—Soft-Fluidic, Soft-Elastic, and Stiff-Elastic—reflecting distinct combinations of stiffness and viscosity. Multiparametric MRI analyses demonstrate that these states map to discrete diffusion- and perfusion-defined microenvironments and localize to specific anatomical tumor compartments, indicating that they capture spatially organized intratumoral variation. Multi-omic analyses using mixed-effects linear models reveal molecular programs specific to each state. In IDH-wildtype glioblastomas, Stiff-Elastic regions correspond to active angiogenic sites with ECM synthesis and remodeling, including vascular collagens, and are enriched for proteins involved in cell contractility and adhesion as well as membrane-stabilizing lipid species, consistent with a mechanically reinforced microenvironment. In contrast, Soft-Elastic regions exhibit loss of fractional anisotropy, reduced vascular permeability, and depletion of both ECM components and membrane-stabilizing lipids. These regions harbor stress-responsive and hypoxia-associated malignant gene expression programs, potentially describing a hypoxic, metabolically constrained niche. Soft-Fluidic regions preferentially engage progenitor-like malignant states, including neuronal lineage programs, and display a glycolytic metabolic profile, consistent with a metabolically active compartment that supports cellular plasticity and turnover. In IDH-mutant gliomas, Stiff-Elastic regions showed transcriptional enrichment for ECM organization without the strong angiogenic or hypoxic signatures, and proteomic changes consistent with altered immune visibility and reduced membrane plasticity. In contrast, Soft-Fluidic and Soft-Elastic regions were enriched for triglycerides and proteins involved in cytoskeleton–membrane coupling. These findings indicate that while intratumoral mechanical states are conserved across glioma subtypes, their molecular composition differs with IDH status. Together, these results establish intratumoral tissue mechanics as a conserved axis of glioma heterogeneity that links noninvasive imaging to specific biophysical microenvironment states. More broadly, this work highlights MRE as a promising tool for probing glioma pathophysiology and provides a framework for incorporating tissue biomechanics into studies of tumor evolution, plasticity, and therapeutic vulnerability. Citation Format: Maksym Zarodniuk, Siri Fløgstad. Svensson, Skarphéðinn Halldórsson, Maria Gomez. Mahiques, Elies Fuster-García, Einar Osland. Vik-Mo, Kyrre Eeg. Emblem, Meenal Datta. Integrative MR elastography and multi-omics identify conserved biomechanical states in adult diffuse gliomas [abstract]. In: Proceedings of the AACR Special Conference in Cancer Research: Brain Cancer; 2026 Mar 23-25; Philadelphia, PA. Philadelphia (PA): AACR; Cancer Res 2026;86(6_Suppl):Abstract nr A059.

Altered glymphatic function is observed for many neurological diseases. Glioma, one of the most common brain cancers, is known to have altered fluid dynamics in terms of edema and blood-brain barrier breakdown, both features potentially impacting the glymphatic function. To study glioma and its fluid dynamics, we propose a flexible mathematical model, including the tumor, the peri-tumoral edema and the healthy tissue. From a mechanical point of view, we consider the brain as a multicompartment porous medium and model both the fluid movement and the clearance of solutes within the brain. Our results indicate that the impairment of the glymphatic system due to glioma growth is two-fold. First, edema resulting from the leakage of fluid at the blood-brain barrier and/or the occlusion of the interstitial fluid exit routes (notably the perivascular spaces) due to migratory tumor cells result in a slight localized increase of pressure, consequently impairing negatively glymphatic clearance. Second, local changes of porosity (i.e. the volume fraction of certain compartments such as perivascular or extracellular spaces), result in a disruption of the transport of solutes in the brain. Our results indicate that an effect similar to the enhanced permeability and retention is obtained using biologically relevant changes of parameter values of our model. Our mathematical model is the first step towards a digital twin for drug or contrast product delivery within the cerebro-spinal fluid directly (e.g. from intrathecal injection) for patients suffering from gliomas.

Diffuse gliomas are malignant brain tumors that grow widespread through the brain. The complex interactions between neoplastic cells and normal tissue, as well as the treatment-induced changes often encountered, make glioma tumor growth modeling challenging. In this paper, we present a novel end-to-end network capable of future predictions of tumor masks and multi-parametric magnetic resonance images (MRI) of how the tumor will look at any future time points for different treatment plans. Our approach is based on cutting-edge diffusion probabilistic models and deep-segmentation neural networks. We included sequential multi-parametric MRI and treatment information as conditioning inputs to guide the generative diffusion process as well as a joint segmentation process. This allows for tumor growth estimates and realistic MRI generation at any given treatment and time point. We trained the model using real-world postoperative longitudinal MRI data with glioma tumor growth trajectories represented as tumor segmentation maps over time. The model demonstrates promising performance across various tasks, including generating high-quality multi-parametric MRI with tumor masks, performing time-series tumor segmentations, and providing uncertainty estimates. Combined with the treatment-aware generated MRI, the tumor growth predictions with uncertainty estimates can provide useful information for clinical decision-making.

PURPOSE:Intrathecal contrast-enhanced magnetic resonance imaging (MRI), which uses a contrast agent as a cerebrospinal fluid (CSF) tracer, is an emerging technique for in vivo imaging of glymphatic function in humans. T1 mapping enables quantification of tracer concentrations; however, methodological limitations persist. This study aimed to enhance the utility of a standard Look-Locker MRI sequence for quantifying tracer concentrations in both brain parenchyma and ventricular CSF. METHODS:Using a 3 T MRI scanner, we conducted (i) optimization of a Look-Locker T1 mapping protocol to determine T1 relaxation times in brain parenchyma and ventricular CSF, (ii) a phantom study to assess repeatability of T1 times estimation and the relaxivity constant r1 calculation, and (iii) a feasibility assessment in patients. RESULTS:An optimized Look-Locker protocol for T1 mapping enabled estimation of T1 relaxation times in brain parenchyma and ventricular CSF that compare with previously reported T1 times. The phantom study demonstrated repeatability of T1 time estimations. In six patients with idiopathic normal pressure hydrocephalus, the method proved feasible for estimating CSF tracer concentrations in brain and ventricular CSF over time following intrathecal MRI contrast injection (gadobutrol, 0.5 mmol). CONCLUSION:Optimizing a standard Look-Locker T1 mapping protocol allowed for estimation of reliable T1 times in both brain parenchyma and ventricular CSF, and the subsequent determination of concentrations of intrathecal CSF tracer in the human brain, supporting its potential for studying glymphatic function. However, limitations with estimation of T1 times and the selection of relaxivity constant of the contrast agent reduce accuracy of concentration estimates.

Glioblastoma (GBM) exhibits two principal growth phenotypes: infiltrative, characterized by diffuse invasion with minimal mass effect, and proliferative, characterized by pronounced tissue compression. Their quantitative delineation and prognostic implications remain uncertain. We introduce an MRI-derived biomarker, the dynamic infiltration rate (DIR), defined as the ratio of tumor-volume expansion to mass-effect–induced peritumoral compression, and evaluate it in silico and clinically. In a synthetic dataset spanning realistic infiltrative-proliferative spectra, DIR correlates strongly with ground truth (R^2=0.85). Applied to patient data, a data-driven threshold separates high- and low-infiltration groups with markedly different overall survival (median 16.0 versus 35.2 weeks; log-rank p<0.001; hazard ratio 2.49). Multivariate Cox analysis adjusted for age, sex, and MGMT status confirms DIR as an independent prognostic factor (HR = 1.38, 95 DIR therefore differentiates proliferative from infiltrative GBM phenotypes and provides prognostic information that could inform personalized therapy and follow-up.

BACKGROUND:Symptomatic meningiomas may require surgical resection to save or improve neurological function. The extent of tumor resection depends on multiple factors, including the tumor's consistency, its location, and the patient's overall condition. This prospective study aims to explore new criteria in combination with previously proposed tumor features on MRI to establish a rapid approach to tumor consistency characterization pre-operatively. METHODS:Forty-eight patients with meningiomas were prospectively included and underwent a dedicated MRI protocol prior to surgery. Qualitative and quantitative MRI characteristics of the tumor were correlated to a previously proposed surgical tumor consistency grading. RESULTS:Soft tumors were associated with homogeneous contrast enhancement, high T2 signal, absence of peritumoral edema (PTE), the presence of tumor cysts, and a uniformly dark appearance on fractional anisotropy (FA) maps. In contrast, firmer tumors were characterized by heterogeneous contrast enhancement, low T2 signal, the presence of PTE, absence of tumor cysts and a heterogeneous appearance on FA maps, requiring supranormal ultrasonic aspirator settings. Tumor signal quantification on T2 and Apparent Diffusion Coefficient maps (ADC) correlated moderately to tumor consistency. T1 sequences did not contribute in determining tumor consistency. CONCLUSION:An array of simple qualitative meningioma characteristics on MRI can assist in swift discrimination of soft and hard tumors preoperatively. These have been displayed in a figure that can easily be implemented clinically for optimal surgical planning.

BackgroundDeep learning‐based segmentation of brain metastases relies on large amounts of fully annotated data by domain experts. Semi‐supervised learning offers potential efficient methods to improve model performance without excessive annotation burden.PurposeThis work tests the viability of semi‐supervision for brain metastases segmentation.Study TypeRetrospective.SubjectsThere were 156, 65, 324, and 200 labeled scans from four institutions and 519 unlabeled scans from a single institution. All subjects included in the study had diagnosed with brain metastases.Field Strength/Sequences1.5 T and 3 T, 2D and 3D T1‐weighted pre‐ and post‐contrast, and fluid‐attenuated inversion recovery (FLAIR).AssessmentThree semi‐supervision methods (mean teacher, cross‐pseudo supervision, and interpolation consistency training) were adapted with the U‐Net architecture. The three semi‐supervised methods were compared to their respective supervised baseline on the full and half‐sized training.Statistical TestsEvaluation was performed on a multinational test set from four different institutions using 5‐fold cross‐validation. Method performance was evaluated by the following: the number of false‐positive predictions, the number of true positive predictions, the 95th Hausdorff distance, and the Dice similarity coefficient (DSC). Significance was tested using a paired samples t test for a single fold, and across all folds within a given cohort.ResultsSemi‐supervision outperformed the supervised baseline for all sites with the best‐performing semi‐supervised method achieved an on average DSC improvement of 6.3% ± 1.6%, 8.2% ± 3.8%, 8.6% ± 2.6%, and 15.4% ± 1.4%, when trained on half the dataset and 3.6% ± 0.7%, 2.0% ± 1.5%, 1.8% ± 5.7%, and 4.7% ± 1.7%, compared to the supervised baseline on four test cohorts. In addition, in three of four datasets, the semi‐supervised training produced equal or better results than the supervised models trained on twice the labeled data.Data ConclusionSemi‐supervised learning allows for improved segmentation performance over the supervised baseline, and the improvement was particularly notable for independent external test sets when trained on small amounts of labeled data.Plain Language SummaryArtificial intelligence requires extensive datasets with large amounts of annotated data from medical experts which can be difficult to acquire due to the large workload. To compensate for this, it is possible to utilize large amounts of un‐annotated clinical data in addition to annotated data. However, this method has not been widely tested for the most common intracranial brain tumor, brain metastases. This study shows that this approach allows for data efficient deep learning models across multiple institutions with different clinical protocols and scanners.Level of Evidence3Technical EfficacyStage 2

BACKGROUND:Cerebrospinal fluid (CSF) serves as a medium for nutrient delivery and waste clearance. The T1 relaxation rate, R1, can be used to measure the concentration of intrinsic solutes and extrinsic contrast agents. PURPOSE:To implement a method for R1 mapping and segmentation of CSF and to use this method to explore how R1 of CSF relates to protein content and gadobutrol after intrathecal administration. STUDY TYPE:Prospective cohort study, complemented by phantom analysis. POPULATION:Ten healthy control subjects (mean age 65.5 ± 4.4 years, range 57-72 years; five males and five females) and protein- and gadobutrol-gradient phantom. FIELD STRENGTH/SEQUENCE:3 T Philips Ingenia scanner; 3D T2W mixed inversion recovery spin-echo (T2W-mixed IRSE) sequence and 3D T1W turbo field echo (3D T1W-TFE). ASSESSMENT:R1 maps were calculated by combining IR and SE data. An automated segmentation method derived from FreeSurfer employed SE data for CSF segmentation and T1W-TFE for anatomical reference. CSF was collected by lumbar puncture for protein measurements, and 0.25 mmol gadobutrol was injected intrathecally. Post-contrast assessments were performed at 3, 24, 48, and 72 h. STATISTICAL TESTS:One-way ANOVA, followed by a post hoc Tukey HSD test, and simple and multiple linear regression analysis; significance level of 0.05. RESULTS:R1 of ventricular CSF 0.216 ± 0.001 s-1 was significantly lower than that surrounding the cerebellum 0.225 ± 0.001 and cerebrum 0.228 ± 0.002 and correlated with lumbar protein concentration (R 2 = 0.56). Peak gadobutrol concentrations were 101 ± 84 μM in ventricles, 185 ± 89 μM in cerebellar SAS and 166 ± 91 μM in cerebral SAS. Corresponding concentrations were 6 ± 4, 17 ± 8, and 37 ± 18 μM at 72 h. DATA CONCLUSION:Intrinsic R1 of CSF in the subarachnoid space correlated with protein content. Intracranial CSF enrichment after intrathecal administration of gadobutrol showed a large variation among healthy volunteers. EVIDENCE LEVEL:2. TECHNICAL EFFICACY:3.

Background Differentiating post-radiation MRI changes from progressive disease (PD) in glioblastoma (GBM) patients represents a major challenge. The clinical problem is two-sided; avoid termination of effective therapy in case of pseudoprogression (PsP) and continuation of ineffective therapy in case of PD. We retrospectively assessed the incidence, management, and prognostic impact of PsP and analyzed factors associated with PsP in a GBM patient cohort. Methods Consecutive GBM patients diagnosed in the South-Eastern Norway Health Region from 2015 to 2018 who had received RT and follow-up MRI were included. Tumor, patient, and treatment characteristics were analyzed in relationship to re-evaluated MRI examinations at 3 and 6 months post-radiation using Response Assessment in Neuro-Oncology criteria. Results A total of 284 patients were included in the study. PsP incidence 3 and 6 months post-radiation was 19.4% and 7.0%, respectively. In adjusted analyses, methylated O6-methylguanine-DNA methyltransferase (MGMT) promoter and the absence of neurological deterioration were associated with PsP at both 3 (p < .001 and p = .029, respectively) and 6 months (p = .045 and p = .034, respectively) post-radiation. For patients retrospectively assessed as PD 3 months post-radiation, there was no survival benefit of treatment change (p = .838). Conclusions PsP incidence was similar to previous reports. In addition to the previously described correlation of methylated MGMT promoter with PsP, we also found that absence of neurological deterioration significantly correlated with PsP. Continuation of temozolomide courses did not seem to compromise survival for patients with PD at 3 months post-radiation; therefore, we recommend continuing adjuvant temozolomide courses in case of inconclusive MRI findings.

BackgroundTo date, multiple advanced magnetic resonance imaging (MRI) methods beyond conventional qualitative structural imaging for the diagnosis, prognosis, and treatment follow-up of glioma have demonstrated their utility for clinical studies. However, these methods often rely on complex off-scanner processing to yield the most information and to extract quantitative biomarkers, limiting their practical use for studies, as well as their clinical translation.While community-driven software solutions exist for these advanced MRI methods, many aspiring clinical researchers face challenges in acquiring the necessary knowledge to effectively apply these tools. This guide, an initiative of the Glioma MR imaging 2.0 network (GliMR), aims to provide an overview of existing solutions, communities, and repositories with the ultimate goal of enabling standardization, open science, and reproducible quantitative imaging studies of gliomas. Yet, most of the reviewed tools and approaches to image data analyses may also be used in the context of studies on diseases other than glioma.ContentThis guide summarizes the state-of-the-art processing software solutions and the repositories/communities for the following advanced MRI methods: DSC; DCE; ASL; diffusion MRI; relaxometry; MRF; MRS; CEST; SWI; QSM; MRE; and task-based and resting-state fMRI. For each of those, after a short introduction about the method and output parameters, the required and recommended image processing steps and quality control measures are described, and we point to further literature for more details. In addition, an overview of openly available software tools that provide these functionalities for MRI processing and exemplify workflows is given. Wherever possible, the readers are guided toward existing inventories, repositories, and communities, which offer not only a collection of these tools, but also more in-depth guidance. Each part concludes with an appraisal of the estimated required expertise and future development needs.ConclusionThis guide provides an extensive overview of the currently available processing tools that can help aspiring clinical researchers to obtain high-quality reproducible imaging data from advanced MRI scans of gliomas. While GliMR n is focused on glioma research, this guide will also be helpful for other clinical neuroimaging topics as general processing steps may not be specific to glioma only.

BACKGROUND:We recently conducted a phase 2 trial (NCT028865685) evaluating intracranial efficacy of pembrolizumab for brain metastases (BM) of diverse histologies. Our study met its primary efficacy endpoint and illustrates that pembrolizumab exerts promising activity in a select group of patients with BM. Given the importance of aberrant vasculature in mediating immunosuppression, we explored the relationship between immune checkpoint inhibitor (ICI) efficacy and vascular architecture in the hopes of identifying potential mechanisms of intracranial ICI response or resistance for BM. METHODS:Using Vessel Architectural Imaging, a histologically validated quantitative metric for in vivo tumor vascular physiology, we analyzed dual-echo DSC/DCE MRI for 44 patients on trial. Tumor and peri-tumor cerebral blood volume/flow, vessel size, arterial and venous dominance, and vascular permeability were measured before and after treatment with pembrolizumab. RESULTS:BM that progressed on ICI were characterized by a highly aberrant vasculature dominated by large-caliber vessels. In contrast, ICI-responsive BM possessed a more structurally balanced vasculature consisting of both small and large vessels, and there was a trend toward a decrease in under-perfused tissue, suggesting a reversal of the negative effects of hypoxia. In the peri-tumor region, the development of smaller blood vessels, consistent with neo-angiogenesis, was associated with tumor growth before radiographic evidence of contrast enhancement on anatomical MRI. CONCLUSIONS:This study, one of the largest functional imaging studies for BM, suggests that vascular architecture is linked with ICI efficacy. Studies identifying modulators of vascular architecture, and effects on immune activity, are warranted and may inform future combination treatments.

Glioblastoma (GBM) is an aggressive brain tumor in which primary therapy is standardized and consists of surgery, radiotherapy (RT), and chemotherapy. However, the optimal time from surgery to start of RT is unknown. A high-grade glioma cancer patient pathway (CPP) was implemented in Norway in 2015 to avoid non-medical delays and regional disparity, and to optimize information flow to patients. This study investigated how CPP affected time to RT after surgery and overall survival. This study included consecutive GBM patients diagnosed in South-Eastern Norway Regional Health Authority from 2006 to 2019 and treated with RT. The pre CPP implementation group constituted patients diagnosed 2006–2014, and the post CPP implementation group constituted patients diagnosed 2016–2019. We evaluated timing of RT and survival in relation to CPP implementation. A total of 1212 patients with GBM were included. CPP implementation was associated with significantly better outcomes (p < 0.001). Median overall survival was 12.9 months. The odds of receiving RT within four weeks after surgery were significantly higher post CPP implementation (p < 0.001). We found no difference in survival dependent on timing of RT below 4, 4–6 or more than 6 weeks (p = 0.349). Prognostic factors for better outcomes in adjusted analyses were female sex (p = 0.005), younger age (p < 0.001), solitary tumors (p = 0.008), gross total resection (p < 0.001), and higher RT dose (p < 0.001). CPP implementation significantly reduced time to start of postoperative RT. Survival was significantly longer in the period after the CPP implementation, however, timing of postoperative RT relative to time of surgery did not impact survival.

This guide summarizes the state-of-the-art processing software solutions and the repositories/communities for the following advanced MRI methods: DSC; DCE; ASL; diffusion MRI; relaxometry; MRF; MRS; CEST; SWI; QSM; MRE; and task-based and resting-state fMRI. For each of those, after a short introduction about the method and output parameters, the required and recommended image processing steps and quality control measures are described, and we point to further literature for more details. In addition, an overview of openly available software tools that provide these functionalities for MRI processing and exemplify workflows is given. Wherever possible, the readers are guided toward existing inventories, repositories, and communities, which offer not only a collection of these tools, but also more in-depth guidance. Each part concludes with an appraisal of the estimated required expertise and future development needs.

Standard treatment of patients with glioblastoma includes surgical resection of the tumor. The extent of resection (EOR) achieved during surgery significantly impacts prognosis and is used to stratify patients in clinical trials. In this study, we developed a U-Net-based deep-learning model to segment contrast-enhancing tumor on post-operative MRI exams taken within 72 h of resection surgery and used these segmentations to classify the EOR as either maximal or submaximal. The model was trained on 122 multiparametric MRI scans from our institution and achieved a mean Dice score of 0.52 ± 0.03 on an external dataset (n = 248), a performance ­on par with the interrater agreement between expert annotators as reported in literature. We obtained an EOR classification precision/recall of 0.72/0.78 on the internal test dataset (n = 462) and 0.90/0.87 on the external dataset. Furthermore, Kaplan-Meier curves were used to compare the overall survival between patients with maximal and submaximal resection in the internal test dataset, as determined by either clinicians or the model. There was no significant difference between the survival predictions using the model's and clinical EOR classification. We find that the proposed segmentation model is capable of reliably classifying the EOR of glioblastoma tumors on early post-operative MRI scans. Moreover, we show that stratification of patients based on the model's predictions offers at least the same prognostic value as when done by clinicians.

Distinctive traits of malignant tumours are abnormal angiogenesis and high pressure. Conventional magnetic resonance imaging (MRI) plays a critical role in radiological evaluation of patients and tumour grading, but challenges remain. Pressure and vasculature have a strong impact on the tissue rheology and therefore they can be quantified by Magnetic Resonance Elastography (MRE). We show that MRE allows to quantify non-invasively tumour grade using pressure and tumour vasculature through wave scattering. We believe that MRE could play a central role in tumour grading and diagnosis as well as in therapy planning and dosage, especially in multidrug treatments scenarios.

Abstract BACKGROUND Radiotherapy remains a cornerstone in glioblastoma (GBM) management. A continued and unmet need in neuro-oncology is efficient non-invasive methods to differentiate post-radiation magnetic resonance imaging (MRI) changes from progressive disease (PD) in GBM patients. The clinical challenge is two-sided; avoid termination of effective therapy in case of pseudoprogression (PsP) and continuation of ineffective therapy in case of PD. We retrospectively assessed incidence, management, prognostic impact, and associated factors with PsP in a real-world GBM patient cohort. MATERIAL AND METHODS Adult (≥18 years) GBM patients diagnosed in the South-Eastern Health Region of Norway from 2015 to 2018, had received radiotherapy, and follow-up MRIs were included. Patient, tumor, and treatment characteristics were analyzed in relationship to re-evaluated MRI examinations at three and six months post-radiation using response assessment in neuro-oncology criteria. RESULTS PsP incidences at three and six months post-radiation were 20% and 8%, respectively. In adjusted analyses, methylated O6-methylguanine-DNA methyltransferase (MGMT) promoter and absence of neurological deterioration were associated with PsP at both three (p<0.001 and p=0.04, respectively) and six months (p=0.003 and p=0.04, respectively) post-radiation. Patients with PsP at MRI three months post-radiation had a median OS of 24.8 months compared to 11.4 months for patients with PD and 18.7 months in patients with stable disease (SD); the difference was significant only when comparing patients with PsP and PD (p<0.001). There was no survival benefit of treatment change for patients retrospectively evaluated as PD three months post-radiation (p=0.8). Median OS for patients in the PsP group was 31.8 months and thus longer than in the PD group (13.1 months) and the SD group (24.1 months). The difference was only significant when comparing patients with PsP and PD (p<0.001). CONCLUSION PsP incidence in this retrospective material was similar to previous reports. In addition to the previously described correlation of methylated MGMT promoter with PsP, we also found that the absence of neurological deterioration significantly correlated to PsP. Continuation of temozolomide courses did not compromise survival for patients with PD at three months post-radiation; therefore, we recommend continuing adjuvant temozolomide courses in case of inconclusive MRI findings. Funding: European Union's Horizon 2020 Programme European Research Council Grant 758657-ImPRESS.

PDF file - 5.2MB, Figure S1: Decreased Flow. Representative example of patient with flow decrease. (A) Anatomic MR imaging showing decrease in the contrast enhanced tumor area. (B) Flow maps showing decreasing flow. The blue ovals indicate region of tumor. (C) Histogram analysis of enhancing tumor showing decrease of flow compared to reference tissue.