Glioblastoma is a highly malignant brain tumor in which maximal safe resection is associated with improved survival, yet the oncological benefit of resection varies by molecular subtype. Recent work has shown that DNA methylation-defined subtypes, particularly receptor tyrosine kinase (RTK) I and II, benefit from complete CE (contrast-enriched) resection compared to mesenchymal tumors, highlighting the need for pre- or intraoperative tools that guide resection based on tumor biology. Here, we present iSTAMP ( i ntraoperative S patially-informed T umor A rchitecture M apping and P rofiling) a real-time, label-free molecular classification framework using stimulated Raman scattering microscopy and graph-based deep learning to predict glioblastoma epigenetic subtypes intraoperatively (within 5-7 minutes). Across 1,295 intraoperative tissue samples from 236 patients profiled with EPIC methylation arrays, our graph attention network achieved high predictive performance for all major subtypes (AUC range 0.88-0.99), with spatially stable predictions across tumor regions. RTK subtypes, but not mesenchymal tumors, showed significant survival benefit from GTR (HR = 0.42, p = 6.1 ×10 -6 ). Explainable AI methods revealed subtype-specific histopathological features, including necrosis and macrophage infiltration in mesenchymal tumors versus glio-fibrillary matrix or axon-rich regions in RTK tumors. Spatial transcriptomic validation confirmed cellular correlates with defined subtype specific SRH features. These findings support the integration of Raman-based molecular diagnostics into intraoperative workflows to guide biologically informed surgical strategies in glioblastoma.

It is assumed that amino acid PET positivity of gliomas corresponds to tracer uptake by tumor cells via the amino acid transporter heterodimer CD98, consisting of a light (LAT1) and heavy subunit (CD98hc), but correlative studies of human tumor tissue samples and PET imaging are largely lacking. Methods: We evaluated a series of 31 patients with diffuse glioma (18 IDH-mutant, 13 IDH-wild type), of whom preoperative amino acid PET images and corresponding tumor tissue samples were available. LAT1 and CD98hc expression were immunohistochemically assessed and correlated to amino acid PET tracer uptake parameters. Results: LAT1 and CD98hc subunits were strongly and almost exclusively expressed on endothelial cells, with most tumor cells lacking immunostaining. There was no correlation of LAT1 or CD98hc expression with amino acid PET uptake parameters. Conclusion: Further studies to clarify the molecular basis of amino acid tracer uptake in diffuse glioma are warranted.

Fluorescence-guided surgery improves intraoperative brain tumor visualization, but currently available agents remain unreliable for several common entities, particularly lower-grade gliomas and lesions without visible 5-aminolevulinic acid (5-ALA)-induced fluorescence. We prospectively analyzed 69 surgically obtained tumor specimens from 61 patients with WHO grade 4 gliomas, WHO grade 2/3 gliomas, meningiomas, and brain metastases before and after ex vivo incubation in 4 µM water-soluble high-load hypericin-polyvinylpyrrolidone complex (HHL-PVP). Fluorescence lifetime and intensity were quantified using a dual-tap CMOS camera system with a hypericin-specific 575-615 nm bandpass filter. HHL-PVP incubation significantly increased both fluorescence intensity and lifetime in all tumor entities, including specimens without visible 5-ALA fluorescence. An independently derived combined lifetime/intensity regression model discriminated pre- from post-incubation measurements with an area under the receiver operating characteristic curve of 0.975, sensitivity of 98.6%, and specificity of 82.6%; entity-specific areas under the curve ranged from 0.933 to 1.000. These findings support robust hypericin-associated signal detection across major brain tumor entities after ex vivo incubation and provide a rationale for future in vivo evaluation of HHL-PVP in fluorescence-guided neuro-oncological surgery.

Dynamic deuterium metabolic imaging (DMI) enables time-resolved mapping of cerebral glucose metabolism in vivo, yet its intrinsically low SNR often renders voxel-wise metabolite quantification unstable-particularly at early repetitions. Low-rank denoising is widely used in MR spectroscopic imaging (MRSI)/DMI to improve robustness, but very low-SNR regimes and dynamic studies remain challenging. Self-supervised learning is promising for DMI/MRSI denoising, but its performance depends strongly on the representation domain and the noise assumptions underlying the training objective Here, we study a pragmatic denoising pipeline for dynamic DMI/MRSI that combines a mild low-rank stabilization with self-supervised learning in the spectral-temporal (f×T) domain. Exploiting metabolite-specific spectral structure and redundancy across repeated measurements, our approach relies on two mild assumptions: (i) additive, approximately zero-mean noise and (ii) approximate noise independence across repeated acquisitions. These conditions are expected to be satisfied across essentially any reconstruction and preprocessing pipeline. Our results indicate that the f×T domain is effective both with spatially correlated and approximately uncorrelated noise. We evaluate the approach in simulations and in vivo dynamic DMI data from healthy volunteers (n=6) and a brain tumor patient. The proposed pipeline improves the robustness of time-resolved metabolite estimates and increases LCModel fit stability relative to a state-of-the-art low-rank baseline (tMPPCA), with the largest gains for weak metabolites and early low-SNR repetitions. Together, this enables more reliable dynamic metabolite mapping in low-SNR regimes.

PURPOSE:Given the early recurrence of brain metastasis (BM), identifying factors that drive BM progression is of clinical interest. This study investigates genetic, epigenetic, and inflammatory signatures in progressive BM following different therapeutic approaches. METHODS:A total of 153 patients who underwent surgical resection for progressive BM were grouped according to the therapeutic strategies prior to the first BM resection: prior radiation (n = 43), systemic therapy (n = 37), combined radiation and systemic treatment (n = 10), and treatment-naive patients (n = 63). Among the treatment-naive patients, 35/63 (55.5%) experienced another intracranial relapse and underwent a second resection (=relapse group), enabling paired analyses. Of these, 23/35 (65.7%) received no therapy between resections; 12/35 (34.3%) received CNS-directed radiotherapy. Tissue samples were analysed using whole-exome sequencing, DNA methylation profiling, and immunohistochemistry. RESULTS:BM resected after progression following prior cranial radiotherapy (43/153, 28.1%) showed significantly lower densities of CD3 + , CD8 + , and CD45RO + cells together with increased FOXP3 + cell density compared with treatment-naïve BM (63/153, 41.2%; median CD3 +: 71 vs. 494 cells/mm²; CD8 +: 44 vs. 187 cells/mm²; CD45RO+: 104 vs. 302 cells/mm²; FOXP3 +: 215 vs. 41 cells/mm²). In the paired analyses, progressive specimen after prior radiation were likewise associated with significantly reduced CD3 + , CD8 + , and CD45RO + and increased FOXP3 + cell densities compared with the matched baseline specimen. In contrast, no genetic alterations or differences in DNA methylation patterns between irradiated and non-irradiated matched samples were identified. CONCLUSION:Progressive BM following cranial radiotherapy demonstrated a distinct immune marker profile consistent with a more immunoregulatory, rather immunosuppressive tumour microenvironment. No therapy-associated genetic or epigenetic alterations were identified. Further prospective studies are warranted to determine whether these immune alterations reflect treatment-related effects or biological features associated with resistance following radiotherapy.

High-grade gliomas (HGGs) are the most aggressive adult brain tumors, with a dismal median survival of approximately 15 months, highlighting the need for novel therapeutic strategies. In a prior immunotherapy trial using dendritic cells against glioblastoma, miR-216b emerged as a potential predictive biomarker. Thus, we hypothesize that miR-216b impacts glioma aggressiveness and thereby therapeutic success. Here, we demonstrate that miR-216b is significantly downregulated in the majority of Isocitrate dehydrogenase 1/2 (IDH) wild-type HGG tissue samples (n = 42) and cell models (n = 18). Functional assays revealed that miR-216b overexpression impairs glioma cell proliferation, migration, and stemness characteristics. Transcriptomic and target prediction analyses identified CDK4, a key cell cycle regulator, as a direct target of miR-216b, confirmed via luciferase reporter assays. Correspondingly, upregulating miR-216b (mimic) via transfection decreased CDK4 mRNA and protein levels accompanied by a p21-dependent increase of cells in G0/G1 phase. In addition, miR-216b expression correlated with increased sensitivity to the CDK4/6 inhibitor Abemaciclib. Notably, miR-216b levels were significantly higher in less aggressive IDH-mutant gliomas (n = 21), linking its downregulation to malignancy grade. Collectively, our findings discovered miR-216b as a tumor suppressor in HGGs, modulating CDK4 expression and affecting the responsiveness to CDK4/6 inhibitors. The observed results support the potential of miR-216b as both a prognostic and predictive indicator in HGGs.

BACKGROUND:Artificial Intelligence (AI) is rapidly emerging as a transformative tool in medical research and practice. In neuro-oncology, AI may help to enhance diagnostic accuracy and reproducibility, manage complex multi-modal data, and facilitate personalized treatment. METHODS:This review aims to provide an overview of AI applications in the analysis of histopathological and molecular data of brain tumors. RESULTS:Key applications in histopathology include molecular biomarker prediction from H&E stained slides, tumor classification, grading, and prognostication. In molecular pathology, the machine learning-driven DNA methylation-based classification of CNS tumors has already become an integral part of the most recent WHO classification. This framework is continuously refined by ongoing research identifying novel tumor types. Two examples of emerging applications are Stimulated Raman Histology (SRH) and nanopore sequencing. SRH enables an intraoperative AI-powered assessment of the histopathological phenotype. Nanopore sequencing can be used for fast molecular profiling of CNS tumors, including intraoperative methylation-based subtyping. Despite these significant advances, the clinical translation of AI tools faces some challenges, including the limited dataset availability, standardization and representativeness; the lack of robust external validation in many published studies; and the limited model interpretability. These challenges are currently being tackled by efforts to compile multi-institutional pathological datasets and by advances in explainable AI. CONCLUSIONS:AI holds promise for advancing personalized neuro-oncology by improving diagnostic accuracy and accelerating existing workflows. Its potential to democratize access to precision diagnostics hinges on efforts to reduce the costs of digital infrastructure and facilitate specialized training.

Glioblastoma, the most frequent and most malign brain cancer, not only cultivates a local immunosuppressive milieu but also causes systemic immunological dynamics. Radiomics is an advanced, automated imaging analysis approach that harnesses data point patterns not readily visible for the human eye. It has been shown that radiomics can differentiate glioblastoma from other tumors, that it can recognize molecular features and that it can identify local immune infiltration in the tumor. However, whether radiomics can also indicate systemic, i.e. peripheral blood, immune states has not been investigated so far. Therefore, we retrospectively analyzed magnetic resonance images of a comprehensively immunophenotyped clinical cohort (n = 34) and performed radiomics feature extraction from three morphological segments of the tumor: the necrotic core, the contrast-enhancing margin and the T2/FLAIR hyperintensive peritumoral zone. 321 radiomics dimensions were then integrated with 67 peripheral blood immunology markers (from flow cytometry and PCR). Via machine learning methods like t-SNE dimensionality reduction and hierarchical clustering, as well as regression modelling, we integrated the highly multidimensional data. A radiomics variable of the T2 hyperintensity zone seemed to predict T helper 17 blood levels. Radiomics variables of the necrotic core were apparently correlated with blood immune cell RORγT levels and CD15 + myeloid cell abundance. Major immune activation parameters like the number of naïve and activated CD8 + T cells, early-differentiated CD8 + T cells, CD56 + natural killer cells or levels of the T helper 1-polarizing transcription factor T-bet could be delineated by integrated multivariable modelling of radiomics features. In an exploratory study on a modestly-sized but immunologically well-characterized glioblastoma cohort we provide first hypothesis-generating evidence that data-driven radiomics approaches could delineate systemic immune states. In the future, non-invasive, radiomics-based blood immunology prediction could e.g. be helpful for patient stratification or immunotherapy research. Before that, however, additional confirmatory studies are needed given the inherent limitations of this work.

We investigated whether metabolic ratios derived from ultra-high-field 7-T 3D-FID-CRT-MRSI can predict intraoperatively visible 5-aminolevulinic acid (5-ALA) fluorescence in gliomas and compared their predictive performance to established imaging markers, including contrast enhancement (CE) on MRI and PET tumor-to-normal ratio (TNR). We retrospectively analyzed 43 patients with histopathologically confirmed adult-type diffuse gliomas (CNS WHO grades 2–4) who underwent preoperative 7-T MRSI and 5-ALA-guided resection. Group differences between 5-ALA-positive and 5-ALA-negative tumors were tested for 16 metabolic ratios to either total creatine (tCr) or combined N-acetylaspartate and N-acetyl-aspartyl-glutamate (NAA + NAAG; total NAA; tNAA) using non-parametric statistics with Šidák correction. CE-MRI status and PET TNR (subcohort, n = 31) were included as reference predictors. We additionally evaluated a subgroup of non-enhancing gliomas (n = 27). Receiver operating characteristic (ROC) analysis was performed to determine diagnostic performance. 5-ALA-positive gliomas demonstrated significantly altered metabolic profiles, showing lower mI/tNAA (p < 0.001) and higher Gln/tCr, Glx/tCr, Gly/tCr, and GSH/tCr ratios (all p < 0.001). These ratios achieved high predictive accuracy for fluorescence (AUCrange = 0.79–0.94), comparable or superior to PET TNR (AUC = 0.90) and CE-MRI (AUC = 0.84). In a subcohort of nonenhancing gliomas, Gly/tCr and Gln/tCr showed a high prediction accuracy (AUC = 0.90). 7-T MRSI metabolic ratios can predict intraoperative 5-ALA fluorescence and may serve as an alternative or adjunct to CE-MRI and PET for preoperative patient selection for 5-ALA administration. Finally, these findings could be especially beneficial in non-enhancing gliomas, where CE-MRI offers limited predictive information. Question Does 7-T MRSI enable preoperative prediction of 5-ALA fluorescence to support patient selection for fluorescence-guided glioma surgery? Findings Several 7-T MRSI metabolic ratios (mI/tNAA, Gln/tCr, Glx/tCr, Gly/tCr and GSH/tCr) robustly predicted 5-ALA fluorescence across glioma subtypes, with diagnostic performance comparable to contrast-enhanced MRI and PET. Clinical relevance Ultra-high-field 7-T MRSI enables noninvasive preoperative prediction of intraoperative 5-ALA fluorescence in gliomas with performance comparable to PET and contrast-enhanced MRI, supporting surgical planning without the need for contrast agents or radiation exposure.

The importance of rapid and accurate histologic analysis of surgical tissue in the operating room has been recognized for over a century. Our standard-of-care intraoperative pathology workflow is based on light microscopy and H&E histology, which is slow, resource-intensive, and lacks real-time digital imaging capabilities. Here, we describe an emerging and innovative method for intraoperative histologic analysis, called Intelligent Histology, that integrates artificial intelligence (AI) with stimulated Raman histology (SRH). SRH is a rapid, label-free, digital imaging method for real-time microscopic tumor tissue analysis. SRH generates high-resolution digital images of surgical specimens within seconds, enabling AI-driven tumor histologic analysis, molecular classification, and tumor infiltration detection. We review the scientific background, clinical translation, and future applications of intelligent histology in tumor neurosurgery. We focus on the major scientific and clinical studies that have demonstrated the transformative potential of intelligent histology across multiple neurosurgical specialties, including neurosurgical oncology, skull base, spine oncology, pediatric tumors, and peripheral nerve tumors. Future directions include the development of AI foundation models through multi-institutional datasets, incorporating clinical and radiologic data for multimodal learning, and predicting patient outcomes. Intelligent histology represents a transformative intraoperative workflow that can reinvent real-time tumor analysis for 21st century neurosurgery.

Abstract Spinal tumor surgery requires rapid tissue diagnosis to guide surgical decisions and further treatment strategies, yet current intraoperative methods are time-intensive and require specialized expertise. No AI systems exist for real-time spinal tumor classification during surgery. We developed SpineXtract, the first AI-powered system for rapid intraoperative spinal tumor diagnosis using stimulated Raman histology (SRH) — a label-free Raman spectromics imaging technique without tissue processing available during surgery. We created a transformer-based classifier optimized for spinal tissue characteristics to identify common tumor types: meningioma, schwannoma, ependymoma, and metastasis. The system was tested in an international, multicenter, simulated, single-arm study using existing SRH datasets (44 patients, 142 slide-images) from three international institutions, with final pathological diagnosis as reference standard. SpineXtract achieved a 92.9% macro-average balanced accuracy (95% CI: 85.5–98.2) within 5 minutes (tumor-specific accuracy range, 84.2–98.6%), while providing quantitative microscopic feedback for granular tissue analysis. Performance remained consistent across institutions (macro balanced accuracy 91.4–92.0%) and outperformed existing brain tumor classifiers by 15.6%. Our results demonstrate clinical applicability, enabling rapid intraoperative diagnosis with performance exceeding current methods, potentially transforming intraoperative diagnostic workflows in spinal tumor surgery.

Background: Resection beyond the contrast-enhancing margin could remove infiltrating glioblastoma but might increase neurological risk. Retrospective comparisons suggest benefit but are vulnerable to anatomical and prognostic selection bias. We assessed pooled randomised evidence comparing supramarginal resection (SMR) with gross total resection (GTR). Methods: We did a post-hoc pooled individual-participant analysis of G-SUMIT and the European trial at four Canadian and 12 European academic neurosurgery centres. Adults with anatomically favourable presumed high-grade glioma or glioblastoma were randomly assigned (1:1) to GTR or SMR. The primary cohort comprised 67 patients with histopathologically confirmed glioblastoma. Outcomes were overall and progression-free survival. Cox models adjusted for age, Karnofsky performance status, and trial; Bayesian analyses estimated posterior probabilities of benefit. The trials are registered with ClinicalTrials.gov, NCT04737577 and NCT04243005. Findings: Between July 1, 2020, and June 30, 2025, 82 patients were randomised; 67 entered the primary cohort (33 GTR and 34 SMR). Median follow-up was 23·6 months. There were 25 deaths and 39 progression-free survival events. For SMR versus GTR, unadjusted hazard ratios were 1·08 (95% CI 0·49–2·38) for death and 0·67 (0·36–1·27) for progression or death; adjusted hazard ratios were 0·79 (0·33–1·89) and 0·43 (0·21–0·87), respectively. Under the data-informed prior, posterior probabilities of benefit were 0·83 for death and 0·97 for progression or death. New neurological deficits occurred in three of 33 patients (9%) assigned to GTR and four of 32 (13%) assigned to SMR; no deaths occurred within 30 days.Interpretation SMR was feasible and associated with a lower adjusted hazard of progression or death; overall-survival benefit was not established. These findings support a definitive event-driven trial but not routine adoption of SMR. Funding: Canadian Institutes of Health Research, Norwegian Cancer Society, and Nordic Cancer Union.

IntroductionMagnetic resonance (MR) imaging is essential for diagnosing central nervous system (CNS) tumors, guiding surgical planning, treatment decisions, and assessing postoperative outcomes and complications. While recent work has advanced automated tumor segmentation and report generation, most efforts have focused on preoperative data, with limited attention to postoperative imaging analysis.MethodsThis study introduces a comprehensive pipeline for standardized postsurgical reporting in CNS tumors. Using the Attention U-Net architecture, segmentation models were trained, independently targeting the preoperative tumor core, non-enhancing tumor core, postoperative contrast-enhancing residual tumor, and resection cavity. In the process, the influence of varying MR sequence combinations was assessed. Additionally, MR sequence classification and tumor type identification for contrast-enhancing lesions were explored using the DenseNet architecture. The models were integrated seamlessly into an automated and standardized reporting pipeline, following the RANO 2.0 guidelines. Training was conducted on multicentric datasets comprising 2000 to 7000 patients, incorporating both private and public data, using a 5-fold cross-validation.ResultsEvaluation included patient-, voxel-, and object-wise metrics, with benchmarking against the latest BraTS challenge results. The segmentation models achieved average voxel-wise Dice scores of 87%, 66%, 70%, and 77% for the tumor core, non-enhancing tumor core, contrast-enhancing residual tumor, and resection cavity, respectively. Classification models reached 99.5% balanced accuracy in MR sequence classification and 80% in tumor type classification.DiscussionThe pipeline presented in this study enables robust, automated segmentation, MR sequence classification, and standardized report generation aligned with RANO 2.0 guidelines, enhancing postoperative evaluation and clinical decision-making. The proposed models and methods were integrated into Raidionics, open-source software platform for CNS tumor analysis, now including a dedicated module for postsurgical analysis.

Background:The neurosurgical workforce has expanded markedly across Europe, often accompanied by declining operative exposure per surgeon. Austria, with one of the highest physician and hospital bed densities in the OECD, provides an important case study to assess whether workforce expansion has translated into proportional service provision and maintained training opportunities. Methods:We performed a retrospective, nationwide analysis of official health statistics from Statistik Austria covering 1997-2023. Data included numbers of practicing neurosurgeons, all specialist physicians, population counts, neurosurgical beds, inpatient stays, and cranial procedures. Absolute and per-capita developments were assessed, and services were related to neurosurgeon density. Statistical analyses comprised Kendall's tau-b, Wilcoxon signed-rank, and Friedman tests. Results:The number of practicing neurosurgeons in Austria increased from 97 in 1997 to 301 in 2023 (+ 210.3%), rising from 1.22 to 3.30 per 100,000 inhabitants (+ 170.5%). Growth in neurosurgeon density significantly outpaced both population growth (+ 14.3%) and the overall increase of specialist physicians (+ 77.4%, p = 0.001). Despite this expansion, absolute service provision showed only negligible to moderate increases (beds + 4.7%, inpatient stays + 28.6%, cranial procedures + 0.1%). Adjusted for workforce size, services per neurosurgeon declined sharply: cranial procedures decreased by -67.8%, inpatient stays by -58.6%, and neurosurgical bed capacity per surgeon by -66.3% (all p < 0.001). Regional disparities were pronounced, with Salzburg reaching 6.51 neurosurgeons per 100,000 while Burgenland registered its first only in 2012 and still shows the nationwide lowest density of 1.00 per 100,000. Conclusion:Austria has experienced rapid workforce growth without a parallel rise in neurosurgical case volume, resulting in declining operative exposure per surgeon. These findings highlight risks for training quality, efficiency, and future competitiveness. Evidence-based workforce planning, structured regulation of training intake, and expansion of outpatient neurosurgical services will be essential to ensure sustainable care and safeguard international standards of neurosurgical education.

Background:Novel approaches to guide personalized treatment in glioblastoma are urgently needed. Given the poor predictive value of genetic biomarkers in glioblastoma, we are conducting a prospective clinical trial to investigate the novel approach of cultivated patient-derived tumor cells (PDCs) for ex vivo drug screening. Methods:In this randomized phase 2 study, we are testing the ability of PDC-based ex vivo drug screening to formulate a personalized recommendation for maintenance treatment in patients with newly diagnosed glioblastoma with unmethylated MGMT promoter after combined radio-chemotherapy. Based on overall survival as the primary endpoint, we plan to include 240 patients (120 per group) to show with a power of 80% that we can increase the median survival from 12 to 17 months (hazard ratio 0.7). Patients will be randomized 1:1 to either the standard group (no drug screening) or the intervention group (drug screening and personalized recommendation for maintenance treatment). In the intervention group, automated drug screening will be performed on PDCs with 28 drugs used for the treatment of solid tumors and hematological malignancies. Based on the cytotoxic activity of these drugs, as quantified by relative viability based on adenosine triphosphate levels, a molecular tumor board will recommend a personalized treatment regimen. Results:The first patient was enrolled in July 2024. Interim analysis of the ATTRACT study (NCT06512311) is expected in late 2027, and final results in 2030. Trial Registration:The ATTRACT trial is registered under the ID NCT06512311 (https://clinicaltrials.gov/study/NCT06512311).

Glioblastoma (GBM) is a high-grade glioma marked by high intratumoral molecular heterogeneity, resistance to therapy and poor prognosis. Leveraging a multisampling approach, we investigated the relation between genomic and physical distance, and whether this information was captured by imaging phenotypes and patient outcomes. We profiled 79 spatially distinct tumor regions from 24 GBM patients (F/M = 0.60; median age = 65.5 years) using deep whole-exome sequencing (>300x coverage). Genomic distance was quantified by pairwise Euclidean distance of somatic mutational profiles and integrated with 3D neuronavigation-based spatial coordinates and progression-free survival (PFS). Two distinct growth patterns emerged: expansive, defined by strong correlation between molecular and spatial distance (RE = 0.6), and stochastic, characterized by molecular divergence uncoupled from spatial proximity (RS = -0.2). High molecular distance correlated with reduced PFS (R = -0.5538, p = 0.026), and stochastic expansion predicted unfavorable outcome (p = 0.035), frequently localized to the frontal lobe. Radiomic analysis from contrast-enhanced T1-weighted MRI revealed that molecular distance was positively correlated with heterogeneous texture features (e.g., GLCM entropy, NGTDM complexity; p < 0.05). Stochastic tumors were enriched for radiomic heterogeneity (e.g.), while expansive tumors displayed homogeneous imaging textures, suggesting convergent phenotypic adaptation in genomically diverse tumors. Functional annotation using neuromaps demonstrated that tumors with high molecular distance exhibited positive correlations with differentiation- and angiogenesis-related metrics (SA_axis, genePC1, CBV), and negative associations with expression of neurotransmitter receptors (acetylcholine, serotonin, dopamine). Stochastic tumors correlated with increased Magnetoencephalography (MEG) derived timescales, implicating higher neural network engagement and spatial fluctuations. These findings suggest a link between type of clonal expansion, radiomic-based structural imaging phenotype, and functional neuroanatomy in GBM, providing a multidimensional framework to interpret tumor behavior and identify clinically relevant radiogenomic signatures.

Further evaluation of HER3 expression in BM of breast and lung cancer.

Differentially methylated CpG sites in HER3+ vs HER3- breast cancer samples (HER2+ cohort)