Search Results (862)

Effective and transparent medical diagnosis relies on accurate and interpretable classification of medical images across multiple modalities. This paper introduces an explainable multi-modal image analysis framework based on a dual-stream architecture that fuses handcrafted descriptors with deep features extracted from a custom MobileNet. Handcrafted descriptors include frequency-domain and texture features, while deep features are summarized using 26 statistical metrics to enhance interpretability. In the fusion stage, complementary features are combined at both the feature and decision levels. Decision-level integration combines calibrated soft voting, weighted voting, and stacking ensembles with optimized classifiers, including decision trees, random forests, gradient boosting, and logistic regression. To further refine performance, a hybrid class-specific feature selection strategy is proposed, combining mutual information, recursive elimination, and random forest importance to select the most discriminative features for each class. This hybrid selection approach eliminates redundancy, improves computational efficiency, and ensures robust classification. Explainability is provided through Local Interpretable Model-Agnostic Explanations, which offer transparent details about the ensemble model’s predictions and link influential handcrafted features to clinically meaningful image characteristics. The framework is validated on three benchmark datasets, i.e., BTTypes (brain MRI), Ultrasound Breast Images, and ACRIMA Retinal Fundus Images, demonstrating generalizability across modalities (MRI, ultrasound, retinal fundus) and disease categories (brain tumor, breast cancer, glaucoma). Full article

Glioblastoma, isocitrate dehydrogenase (IDH1/2) wild-type (IDH-wildtype), is one of the most aggressive and malignant tumors of the central nervous system, characterized by rapid growth, pronounced cellular heterogeneity, and an exceptionally poor prognosis. The median survival time for patients with glioblastoma, IDH-wildtype, is approximately 15 months after diagnosis, and current multimodal treatment strategies remain largely ineffective. This review focuses on contemporary pharmacotherapeutic approaches used in the management of glioblastoma, IDH-wildtype, including temozolomide-based chemotherapy, corticosteroids for edema control, and antiangiogenic therapy in recurrent disease, with particular emphasis on their clinical efficacy and limitations. In addition, the review discusses emerging targeted therapeutic strategies developed for IDH-mutant diffuse gliomas, which represent a biologically distinct disease entity. Particular attention is given to ivosidenib, a selective inhibitor of mutant IDH1, currently evaluated for the treatment of astrocytoma, IDH-mutant, grade 4. Its epigenetic mechanism of action, involving inhibition of the oncometabolite 2-hydroxyglutarate (2-HG), is outlined, along with preliminary clinical evidence suggesting potential to delay disease progression. Finally, innovative drug-delivery technologies designed to overcome the blood–brain barrier are briefly discussed as complementary strategies that may enhance the efficacy of both conventional and targeted therapies. Overall, future advances in the treatment of diffuse gliomas will likely depend on the integration of molecularly targeted agents, predictive biomarkers, and advanced delivery platforms aimed at improving patient survival and quality of life. Full article

This work presents the development and evaluation of Artificial Intelligence (AI) models for the automatic classification of brain tumors in Magnetic Resonance Imaging (MRI) scans. Several deep learning architectures were implemented and compared, including VGG-19, ResNet50, EfficientNetB3, Xception, MobileNetV2, DenseNet201, InceptionV3, Vision Transformer (ViT), and an Ensemble model. The models were developed in Python (version 3.12.4) using the Keras and TensorFlow frameworks and trained on a public Brain Tumor MRI dataset containing 7023 images. Data augmentation and hyperparameter optimization techniques were applied to improve model generalization. The results showed high classification performance, with accuracies ranging from 89.47% to 98.17%. The Vision Transformer achieved the best performance, reaching 98.17% accuracy, outperforming traditional Convolutional Neural Network (CNN) architectures. Explainable AI (XAI) methods Grad-CAM, LIME, and Occlusion Sensitivity were employed to assess model interpretability, showing that the models predominantly focused on tumor regions. The proposed approach demonstrated the effectiveness of AI-based systems in supporting early diagnosis of brain tumors, reducing analysis time and assisting healthcare professionals. Full article

IoT-based blockchain technology has improved the healthcare system to ensure the privacy and security of healthcare data. A Blockchain Bridge (BB) is a tool that enables multiple blockchain networks to communicate with each other. The existing approach combining the classical and quantum blockchain models failed to secure the data transmission during cross-chain communication. Thus, this study proposes a new BB verification for secure healthcare data transfer. Additionally, a brain tumor analysis framework is developed based on segmentation and neural networks. After the patient’s registration on the blockchain network, Brain Magnetic Resonance Imaging (MRI) data is encrypted using Hash-Keyed Quantum Cryptography and verified using a Peer-to-Peer Exchange model. The Brain MRI is preprocessed for brain tumor detection using the Fuzzy HaMan C-Means (FHMCM) segmentation technique. The features are extracted from the segmented image and classified using the LeCun Kaiming-based Convolutional ModSwish Neural Network (LCK-CMSNN) classifier. Subsequently, the brain tumor diagnosis report is securely transferred to the patient via a smart contract. The proposed model verified BB with a Verification Time (VT) of 12,541 ms, secured the input with a Security level (SL) of 98.23%, and classified the brain tumor with 99.15% accuracy, thus showing better performance than the existing models. Full article

With the increasing prevalence of brain tumors, it becomes crucial to ensure fast and reliable segmentation in MRI scans. Medical professionals struggle with manual tumor segmentation due to its exhausting and time-consuming nature. Automated segmentation speeds up decision-making and diagnosis; however, achieving an optimal balance between accuracy and computational cost remains a significant challenge. In many cases, current methods trade speed for accuracy, or vice versa, consuming substantial computing power and making them difficult to use on devices with limited resources. To address this issue, we present PRA-UNet, a lightweight deep learning model optimized for fast and accurate 2D brain tumor segmentation. Using a single 2D input, the architecture processes four types of MRI scans (FLAIR, T1, T1c, and T2). The encoder uses inverted residual blocks and bottleneck residual blocks to capture features at different scales effectively. The Convolutional Block Attention Module (CBAM) and the Spatial Attention Module (SAM) improve the bridge and skip connections by refining feature maps and making it easier to detect and localize brain tumors. The decoder uses depthwise separable convolutions, which significantly reduce computational costs without degrading accuracy. The BraTS2020 dataset shows that PRA-UNet achieves a Dice score of 95.71%, an accuracy of 99.61%, and a processing speed of 60 ms per image, enabling real-time analysis. PRA-UNet outperforms other models in segmentation while requiring less computing power, suggesting it could be suitable for deployment on lightweight edge devices in clinical settings. Its speed and reliability enable radiologists to diagnose tumors quickly and accurately, enhancing practical medical applications. Full article

The integration of molecular features into histopathological diagnoses has become central to the World Health Organization (WHO) classification of central nervous system (CNS) tumors, improving prognostic accuracy and supporting precision medicine. However, unequal access to molecular testing limits the universal application of integrated diagnosis. To address this, artificial intelligence (AI) models are being developed to predict molecular alterations directly from histological data. In gliomas, deep learning applied to whole-slide images (WSIs) of permanent sections achieves neuropathologist-level accuracy in predicting biomarkers such as IDH mutation and 1p/19q co-deletion, as well as in molecular subtype classification and outcome prediction. Recent advances extend these approaches to intraoperative cryosections, enabling real-time glioma grading, molecular prediction, and label-free tissue analysis using modalities such as stimulated Raman histology and domain-adaptive image translation. Beyond gliomas, AI-powered histology is being explored in other brain tumors, including morphology-based molecular classification of spinal cord ependymomas and intraoperative discrimination of gliomas from primary CNS lymphomas. This review summarizes current progress in AI-assisted molecular profiling prediction of brain tumors from tissue, highlighting opportunities for rapid, accurate, and globally accessible diagnostics. The integration of histology and computational methods holds promise for the development of smart AI-assisted neuro-oncology. Full article

Lung adenocarcinoma (LUAD), the most prevalent subtype of non-small cell lung carcinoma (NSCLC), commonly metastasizes to the brain, particularly in advanced stages. Since brain metastases (BMs) are a leading cause of morbidity and mortality in LUAD patients, their early detection is critical, necessitating the identification of reliable biomarkers. Gangliosides (GGs), a class of bioactive glycosphingolipids involved in cell signaling, adhesion, and immune regulation, have emerged as promising candidates for diagnostic and therapeutic targeting in LUAD-associated brain metastases (BMLA). In this context, ion mobility spectrometry mass spectrometry (IMS-MS) was employed here to analyze GG alterations in BMLA tissues compared to healthy cerebellar control. The results revealed marked differences, including a reduction in the total number of species, altered sialylation profiles, and variations in fatty acid chain length and sphingoid base hydroxylation. GM3, a monosialodihexosylganglioside, was significantly overexpressed in BMLA, supporting its role in tumor progression via immune evasion and oncogenic signaling. Elevated levels of the brain-specific GT1 ganglioside further point to its possible role as a metastasis-associated biomarker, while the presence of asialogangliosides, absent in normal brain, suggests adaptation to the brain microenvironment. Structural modifications such as O-acetylation, fucosylation, and CH₃COO⁻ were more frequent in BMLA, being associated with aggressive tumor phenotypes. Ceramide profiles revealed increased levels of proliferative C16- and C24-ceramides and decreased pro-apoptotic C18-ceramide. Additionally, GM3(d18:1/22:0) and GD3(d18:1/16:0), identified as potential BMLA biomarkers, were structurally characterized using (−) nanoelectrospray ionization (nanoESI) IMS collision-induced dissociation tandem MS (CID MS/MS). Collectively, these findings highlight the clinical potential of GGs for early diagnosis and targeted therapy in BMLA. Full article

Brain tumors, both malignant and non-malignant, represent a persistent global health challenge. Differences in diagnosis, treatment, and outcomes are influenced by race, ethnicity, socioeconomic status (SES), and geographical location. Brain and central nervous system (CNS) tumors rank 19th in global cancer incidence and 12th in cancer-related mortality. U.S. Incidence is higher in females and individuals with greater socioeconomic means, contrasting with global patterns where males are more affected. Glioblastoma has a wide variation in incidence and survival by state, with rural regions showing higher mortality despite lower incidence, often due to reduced access to specialized care. Non-Hispanic Black children with CNS tumors experience higher mortality than their White peers, even after adjusting for SES. Outcomes are generally poorer in low- and middle-income countries, where healthcare infrastructure remains limited. Biological and genetic differences may also influence treatment response and tumor behavior across population groups. This review outlines key variations in brain tumor care, with a key focus on the United States, and emphasizes the need for patient-centered strategies to ensure timely diagnosis, consistent treatment, and improved outcomes. Full article

Background and Clinical Significance: Tumefactive demyelinating lesions (TDLs) are large demyelinating lesions that mimic intracranial tumors, posing a diagnostic challenge in both clinical presentation and conventional imaging. Distinguishing TDLs from central nervous system tumors can be challenging due to their similar imaging appearances. Specific magnetic resonance imaging (MRI) features such as open-ring contrast enhancement, mild mass effect, lack of cortical involvement, and rapid responsiveness to corticosteroids favor a demyelinating etiology of the lesion. This report presents a case of a tumefactive demyelination lesion showing a T2/fluid-attenuated inversion recovery (FLAIR) mismatch sign suggestive of a low-grade astrocytoma, focusing on imaging findings, therapeutic response, and diagnostic considerations. Case Description: A 63-year-old woman presented with headache, progressive speech impairment, and difficulty swallowing. MRI revealed a large lesion in the left frontal lobe with a T2/FLAIR mismatch sign, which initially suggested a low-grade astrocytoma. Additionally, the lesion was hypodense on noncontrast computed tomography (CT), did not show open-ring enhancement, and only had mild mass effect with perifocal edema. Given these conflicting imaging findings, a biopsy was considered; however, the patient declined the procedure and agreed to a follow-up. Corticosteroid therapy was initiated to reduce swelling, resulting in a significant reduction in the lesion within two weeks. A follow-up MRI confirmed near-complete regression of the lesion after two months. Conclusions: While a T2/FLAIR mismatch sign correlates with isocitrate dehydrogenase (IDH)-mutant 1p/19q non-codeleted astrocytoma, the dynamic radiological and clinical response to corticosteroids was more indicative of demyelination. This case highlights the importance of considering TDLs in the differential diagnosis of tumor-like brain lesions to avoid unnecessary invasive interventions like biopsy or surgical removal. Full article

Background/Objectives: Ependymomas are central nervous system (CNS) tumors with marked biological and clinical heterogeneity, particularly in pediatric populations. While the 2021 World Health Organization (WHO) classification emphasizes molecular subgroups—posterior fossa type A (PFA) and B (PFB), supratentorial ZFTA fusion-positive (ST-ZFTA), and YAP1 fusion-positive (ST-YAP)— routine diagnosis is still based on histology and immunohistochemistry (IHC). Recent single-cell RNA sequencing and spatial transcriptomic studies have revealed distinct tumor cell populations, including ependymal-like, astroglial-like, progenitor-like, and stress-associated states. However, a major unresolved issue is whether such heterogeneity can be appreciated and interpreted on conventional pathology slides. Methods: This study examined ependymomas from the Children’s Brain Tumor Network (CBTN), with hematoxylin and eosin (H&E) and IHC for glial fibrillary acidic protein (GFAP) and epithelial membrane antigen (EMA). Tumor regions were stratified into high-cellularity and low-cellularity regions, and staining patterns were correlated with known cellular features from the prior literature. Results: Low-cellularity zones exhibit strong fibrillary GFAP, resembling astroglial or subependymal differentiation. In contrast, high-cellularity zones more often demonstrate variable EMA patterns and GFAP/EMA-negative compartments, consistent with undifferentiated progenitor-like populations. Perinecrotic areas showed increased GFAP and EMA, possibly reflecting stress-associated cellular states and mesenchymal differentiation. Comparisons between PFA and ST-ZFTA tumors revealed that ST-ZFTA ependymomas were significantly more likely to be hypercellular, with a higher frequency of diffuse EMA expression. In contrast, PFA tumors displayed broader variability with stronger GFAP perinuclear staining. Conclusions: These findings support the concept that conventional histology can capture relevant heterogeneity and may complement molecular studies. The recognition of such features may help refine histopathological assessment and provide practical prognostic insights, particularly in resource-limited settings where molecular testing is not universally available. Full article

Brain tumors encompass a heterogeneous group of neoplasms, including primary and secondary metastatic lesions, with glioblastoma multiforme (GBM) representing the most aggressive primary malignancy. Despite advancements in surgical resection, radiotherapy, and chemotherapy, the prognosis for GBM remains poor due to its infiltrative nature, tumor heterogeneity and resistance mechanisms. Emerging diagnostic tools, such as liquid biopsies, and therapeutic strategies leveraging extracellular vesicles (EVs) are reshaping the field of neuro-oncology. EVs, lipid bilayer-enclosed particles secreted by cells, carry oncogenic cargo such as microRNAs and molecular chaperones, influencing tumor progression, immune evasion, and therapy resistance. Recent research highlights their potential as biomarkers for early diagnosis and vehicles for targeted drug delivery across the blood–brain barrier (BBB). EV-based nanotherapeutics show promise in improving treatment precision, reducing systemic toxicity, and advancing precision medicine in brain tumor management. However, challenges related to EV heterogeneity, cargo-loading efficiency, and large-scale production must be addressed to fully realize their therapeutic potential. This review explores the multifaceted roles of EVs in brain tumors, emphasizing their diagnostic, prognostic, and therapeutic applications. Full article

Purpose: The aim of this review is to evaluate the role of intraoperative flow cytometry (IFC) in tumor surgery. Methods: The Medline, Scopus, and Cochrane databases were searched up to 21 June 2025 to identify all available studies that met the inclusion criteria for final evaluation. To assess the risk of bias and applicability concerns, the Quality Assessment of Diagnostic Accuracy Studies 2 (QUADAS-2) tool was used. Results: A total of 22 studies involving 1511 patients with various tumor types were analyzed to assess the utility of IFC in the rapid diagnosis of tumors. The studies investigated IFC’s role in tumor grading, margin delineation, prognostic evaluation, and in differentiating neoplastic from benign lesions, as well as normal from cancerous tissues. In brain tumors, particularly gliomas and meningiomas, IFC demonstrated high diagnostic performance, with reported sensitivities ranging from 61% to 100% and specificities from 66% to 100%. Studies on non-brain tumors also showed high accuracy in distinguishing neoplastic from normal tissues, with sensitivities and specificities exceeding 85% in most cases. The most promising results were observed in brain tumor surgery, although its application in other tumor types continues to expand. Conclusions: IFC appears to be a valuable intraoperative tool in surgical oncology, providing rapid results within minutes and assisting in surgical and therapeutic decision-making. Nonetheless, studies remain limited, and further research is needed, particularly for non-brain tumors, to establish standardized cut-off values and enhance diagnostic reliability. Full article

Glioblastoma (GBM), the most common primary malignant brain tumor in adults, has a median survival of 14–15 months despite aggressive treatment. Monitoring relies on MRI, but differentiating tumor progression from pseudo-progression or radiation necrosis remains difficult. Plasma extracellular vesicles (EVs) are emerging as promising non-invasive biomarkers due to their molecular cargos and accessibility. This review evaluates studies that specifically isolated plasma EVs for molecular profiling in GBM diagnosis and monitoring. Biomarkers (miRNA, RNA, DNA, proteins), EV characterization methods, and advancements in enriching tumor-derived EV subpopulations and assessing their diagnostic and prognostic potential are highlighted. Plasma EVs carry diverse cargos, including miRNAs (e.g., miR-21, miR-15b-3p), mRNAs (e.g., EGFRvIII), circRNAs, and proteins (e.g., CD44, GFAP). Composite molecular signatures have achieved sensitivities of 87–100% and specificities of 73–100% for GBM diagnosis. Tumor-derived EVs, enriched using techniques like SEC-CD44 immunoprecipitation, microfluidic platforms, or 5-ALA-induced PpIX fluorescence, enhance biomarker detection. Non-tumor-derived EVs may also reflect GBM’s systemic effects. Challenges include EV heterogeneity, non-EV contamination, and variable biomarker expression across studies. Plasma-EV-based liquid biopsies offer significant potential for GBM monitoring, with advanced enrichment methods improving tumor-specific biomarker detection. Standardizing isolation protocols and validating biomarkers in larger cohorts are critical for clinical translation. Full article

Accurate classification of brain tumors from MRI scans remains challenging due to limited annotated data. This study compares data-efficient paradigms—few-shot learning (FSL) and zero-shot learning (ZSL)—for tumor diagnosis using deep learning and vision–language models. A Prototypical Network (ProtoNet) with CNN, ResNet-18, and vision transformer backbones was evaluated under 1000 randomly sampled five-shot, four-way episodes (mean ± SD). The ResNet-18 ProtoNet achieved 85% ± 8% accuracy (F1 = 0.85), surpassing a fine-tuned ResNet-50 baseline (42% ± 12%) and the CLIP (ZSL) model (30% ± 10%). A visual-only ZSL baseline without text guidance achieved 54% ± 11%. These results highlight that metric-based FSL offers 43% absolute improvement over standard fine-tuning and establishes a robust benchmark for data-efficient MRI classification under severe label constraints. Full article

Background: Central nervous system (CNS) tumors during pregnancy are rare but present significant diagnostic, therapeutic, and ethical challenges. These include both primary and metastatic lesions, which share overlapping clinical features and management complexities. Their clinical course is influenced by gestational physiological changes, which can mask symptoms and delay diagnosis, thereby increasing maternal and fetal risks. Objective: This review aims to synthesize current evidence on the epidemiology, pathophysiology, clinical presentation, diagnostic strategies, treatment options, prognosis, and ethical considerations related to CNS tumors in pregnant patients. Methods: A comprehensive literature review was conducted, including retrospective and prospective studies, clinical guidelines, and systematic reviews focusing on brain and spinal tumors diagnosed during pregnancy. Particular attention was given to the impact of gestational age, tumor histology, and maternal condition on treatment outcomes. Results: Hormone-sensitive tumors such as meningiomas and prolactinomas may exhibit accelerated growth during pregnancy due to elevated progesterone and prolactin levels. Diagnosis is often delayed due to symptom overlap with normal gestational changes. MRI without contrast remains the imaging modality of choice. Glucocorticoids and selected chemotherapy agents can be cautiously used depending on gestational age. Surgical resection, particularly in the second trimester, has been shown to be safe and effective in appropriate clinical scenarios. Multidisciplinary coordination is essential. Prognosis varies based on tumor type and timing of intervention, with maternal survival prioritized in high-risk situations. Ethical management hinges on patient autonomy, informed consent, and proportionality of medical interventions. Conclusions: CNS tumors during pregnancy require early recognition, individualized treatment planning, and ethical vigilance. Multidisciplinary collaboration is vital to optimizing outcomes for both mother and fetus. Future efforts should focus on developing standardized protocols and expanding evidence through multicenter studies. Full article