Search Results (173)

User authentication is crucial for safeguarding access to digital systems and services. Biometric authentication serves as a strong and user-friendly alternative to conventional security methods such as passwords and PINs, which are often susceptible to breaches. This study proposes a deep learning-based multimodal biometric system that combines fingerprint (FP) and finger vein (FV) modalities to improve accuracy and security. The system explores three fusion strategies: feature-level fusion (combining feature vectors from each modality), score-level fusion (integrating prediction scores from each modality), and a hybrid approach that leverages both feature and score information. The implementation involved five pretrained convolutional neural network (CNN) models: two unimodal (FP-only and FV-only) and three multimodal models corresponding to each fusion strategy. The models were assessed using the NUPT-FPV dataset, which consists of 33,600 images collected from 140 subjects with a dual-mode acquisition device in varied environmental conditions. The results indicate that the hybrid-level fusion with a dominant score weight (0.7 score, 0.3 feature) achieved the highest accuracy (99.79%) and the lowest equal error rate (EER = 0.0018), demonstrating superior robustness. Overall, the results demonstrate that integrating deep learning with multimodal fusion is highly effective for advancing scalable and accurate biometric authentication solutions suitable for real-world deployments. Full article

Background/Objectives: Multimodal large language models (LLMs) are increasingly used in radiology. However, their ability to recognize fundamental imaging features, including modality, anatomical region, imaging plane, contrast-enhancement status, and particularly specific magnetic resonance imaging (MRI) sequences, remains underexplored. This study aims to evaluate and compare the performance of three advanced multimodal LLMs (ChatGPT-4o, Claude 4 Opus, and Gemini 2.5 Pro) in classifying brain MRI sequences. Methods: A total of 130 brain MRI images from adult patients without pathological findings were used, representing 13 standard MRI series. Models were tested using zero-shot prompts for identifying modality, anatomical region, imaging plane, contrast-enhancement status, and MRI sequence. Accuracy was calculated, and differences among models were analyzed using Cochran’s Q test and McNemar test with Bonferroni correction. Results: ChatGPT-4o and Gemini 2.5 Pro achieved 100% accuracy in identifying the imaging plane and 98.46% in identifying contrast-enhancement status. MRI sequence classification accuracy was 97.7% for ChatGPT-4o, 93.1% for Gemini 2.5 Pro, and 73.1% for Claude 4 Opus (p < 0.001). The most frequent misclassifications involved fluid-attenuated inversion recovery (FLAIR) sequences, often misclassified as T1-weighted or diffusion-weighted sequences. Claude 4 Opus showed lower accuracy in susceptibility-weighted imaging (SWI) and apparent diffusion coefficient (ADC) sequences. Gemini 2.5 Pro exhibited occasional hallucinations, including irrelevant clinical details such as “hypoglycemia” and “Susac syndrome.” Conclusions: Multimodal LLMs demonstrate high accuracy in basic MRI recognition tasks but vary significantly in specific sequence classification tasks. Hallucinations emphasize caution in clinical use, underlining the need for validation, transparency, and expert oversight. Full article

In recent years, cross-domain few-shot learning (CDFSL) has demonstrated remarkable performance in hyperspectral image classification (HSIC), partially alleviating the distribution shift problem. However, most domain adaptation methods rely on similarity metrics to establish cross-domain class matching, making it difficult to simultaneously account for intra-class sample size variations and inherent inter-class differences. To address this problem, existing studies have introduced a class weighting mechanism within the prototype network framework, determining class weights by calculating inter-sample similarity through distance metrics. However, this method suffers from a dual limitation: susceptibility to noise interference and insufficient capacity to capture global class variations, which may lead to distorted weight allocation and consequently result in alignment bias. To solve these issues, we propose a novel class-discrepancy dynamic weighting-based cross-domain FSL (CDDW-CFSL) framework. It integrates three key components: (1) the class-weighted domain adaptation (CWDA) method dynamically measures cross-domain distribution shifts using global class mean discrepancies. It employs discrepancy-sensitive weighting to strengthen the alignment of critical categories, enabling accurate domain adaptation while maintaining feature topology; (2) the class mean refinement (CMR) method incorporates class covariance distance to compute distribution discrepancies between support set samples and class prototypes, enabling the precise capture of cross-domain feature internal structures; (3) a novel multi-dimensional feature extractor that captures both local spatial details and continuous spectral characteristics simultaneously, facilitating deep cross-dimensional feature fusion. The results in three publicly available HSIC datasets show the effectiveness of the CDDW-CFSL. Full article

Background: Acute necrotising encephalopathy (ANE) is a rare and severe type of encephalopathy with bilateral symmetrical brain lesions, often following a viral prodrome. ANE type 1 (ANE1) is a disease subtype with a predisposing mutation in the gene encoding RAN binding protein 2 (RANBP2). Methods: We report a case of a 3-year-old girl with clinical symptoms of ANE and brain MRI findings suggesting ANE1, which was subsequently confirmed by genetic analysis. Results: MRI of the brain demonstrated symmetrical high T2/FLAIR signal changes in the lateral geniculate bodies, claustrum, ventromedial thalami, subthalamic nuclei, mamillary bodies, and brainstem, with partly corresponding diffusion restriction, as well as additional haemorrhagic changes in the lateral geniculate bodies on susceptibility weighted imaging. Genetic analysis revealed a heterozygous pathogenic variant of the RANBP2 gene. With immunosuppressive and supportive treatment, the patient fully recovered and was discharged after 10 days in the hospital with no residual symptoms. Conclusions: Recognition of the characteristic MRI findings in ANE1 can facilitate a timely diagnosis and enhance the clinical management of the patient and their relatives, especially given the high risk of disease recurrence. Full article

Background: The asymmetry of cortical veins in susceptibility weighted imaging (SWI) in MRI might be a biomarker for migraine auras and cortical spreading depression (CSD). The aim of this study was to assess in humans if SWI asymmetry can be found in patients who have migraine attacks without auras. Methods: We included patients (n = 100 per group) from the emergency room setting when they (i) presented with an acute neurological deficit or headache; (ii) had a discharge diagnosis of a migraine aura, a migraine without an aura, or neither (controls without stroke or epilepsy); and (iii) had a brain MRI with SWI in the acute setting. Results: In the migraines with auras group, SWI asymmetry was found in 26% (95% CI 18–35) compared to patients with migraines without auras (3%, [95% CI 1–8], p < 0.001) and controls 7% [95% CI 3–14], p < 0.001). There was no difference between patients with migraines without auras and controls (p = 0.19). Conclusions: The distinct SWI changes in migraines with and without auras suggest that CSD might not be involved in the pathophysiology of migraines without auras. Full article

MRI perfusion is used to diagnose and monitor neurological conditions such as brain tumors, stroke, dementia, and traumatic brain injury. Dynamic Susceptibility Contrast (DSC) is the most widely available quantitative MRI technique for perfusion imaging. Even in its most basic implementation, DSC MRI provides critical hemodynamic metrics like cerebral blood flow (CBF), blood volume (CBV), mean transit time (MTT), and time between the peak of arterial input and residue function (Tmax), through the dynamic tracking of a gadolinium-based contrast agent. Notwithstanding its high clinical importance and widespread use, the reproducibility and diagnostic reliability are impeded by a lack of standardized pre-processing protocols and quality controls. A comprehensive literature review and the authors’ aggregated experience identified common DSC MRI artefacts and corresponding pre-processing methods. Pre-processing methods to correct for artefacts were evaluated for their practical applicability and validation status. A consensus on the pre-processing was established by a multidisciplinary team of experts. Acquisition-related artefacts include geometric distortions, slice timing misalignment, and physiological noise. Intrinsic artefacts include motion, B₁ inhomogeneities, Gibbs ringing, and noise. Motion can be mitigated using rigid-body alignment, but methods for addressing B₁ inhomogeneities, Gibbs ringing, and noise remain underexplored for DSC MRI. Pre-processing of DSC MRI is critical for reliable diagnostics and research. While robust methods exist for correcting geometric distortions, motion, and slice timing issues, further validation is needed for methods addressing B₁ inhomogeneities, Gibbs ringing, and noise. Implementing adequate mitigation methods for these artefacts could enhance reproducibility and diagnostic accuracy, supporting the growing reliance on DSC MRI in neurological imaging. Finally, we emphasize the crucial importance of pre-scan quality assurance with phantom scans. Full article

Background/Objectives: Despite extensive research, the underlying causes of amyotrophic lateral sclerosis (ALS) remain unclear. This umbrella review aims to synthesize a vast body of evidence from advanced magnetic resonance imaging (MRI) studies of ALS, encompassing a wide range of neuroimaging techniques and patient cohorts. Methods: Following the PRISMA guidelines, we conducted an extensive search of four databases (PubMed, Scopus, Web of Science, and Embase) for articles published until 3 December 2024. Data extraction and quality assessment were independently performed using the AMSTAR2 tool. Results: This review included 18 studies that incorporated data from over 29,000 ALS patients. Structural MRI consistently showed gray matter atrophy in the motor and extra-motor regions, with significant white matter (WM) atrophy in the corticospinal tract and corpus callosum. Magnetic resonance spectroscopy revealed metabolic disruptions, including reduced N-acetylaspartate and elevated choline levels. Functional MRI studies have demonstrated altered brain activation patterns and functional connectivity, reflecting compensatory mechanisms and neurodegeneration. fMRI also demonstrated disrupted motor network connectivity and alterations in the default mode network. Diffusion MRI highlighted microstructural changes, particularly reduced fractional anisotropy in the WM tracts. Susceptibility-weighted imaging and quantitative susceptibility mapping revealed iron accumulation in the motor cortex and non-motor regions. Perfusion MRI indicated hypoperfusion in regions associated with cognitive impairment. Conclusions: Multiparametric MRI consistently highlights widespread structural, functional, and metabolic changes in ALS, reflecting neurodegeneration and compensatory mechanisms. Full article

Deep learning has significantly advanced image classification by leveraging hierarchical feature representations. A key factor in enhancing classification accuracy is feature concatenation, which integrates diverse feature sets to provide a richer representation of input data. However, this fusion strategy has inherent limitations, including increased computational complexity, susceptibility to redundant or irrelevant features, and challenges in optimally weighting different feature contributions. To address these challenges, this paper presents a pseudo-multiview learning method that dynamically combines different views at the evidence level using a belief-based model known as subjective logic. This approach adaptively assigns confidence levels to each view, ensuring more effective integration of complementary information while mitigating the impact of noisy or less relevant features. Experimental evaluations of datasets from various domains demonstrate that the proposed method enhances classification accuracy and robustness compared with conventional classification techniques. Full article

Background/Objectives: This study evaluates intratumoral susceptibility signals (ITSS) as imaging markers for glioma grade prediction and their association with molecular and histopathologic features, in the context of the fifth edition of the World Health Organization Classification of Tumors of the Central Nervous System (WHO CNS5). Methods: We retrospectively analyzed patients with adult diffuse gliomas who underwent pretreatment magnetic resonance imaging. ITSS were semiquantitatively graded by two radiologists: grade 0 (no signal), grade 1 (1–5), grade 2 (6–10), and grade 3 (≥11). Relative cerebral blood volume (rCBV) and tumor volume were also obtained. Histopathologic features included tumor grade, Ki-67, mitotic count, necrosis, microvascular proliferation, and molecular alterations (isocitrate dehydrogenase [IDH], 1p/19q, cyclin-dependent kinase inhibitors 2A and 2B [CDKN2A/B], and p53). Regression models predicted tumor grade (low: 1–2, high: 3–4) using ITSS, tumor volume, and rCBV. ROC curves and diagnostic performance metrics were analyzed. Results: 99 patients were included. ITSS grading correlated with rCBV, tumor volume, mitotic count, Ki-67, and tumor grade (p < 0.001). ITSS grades 0–1 were associated with oligodendrogliomas and astrocytomas (p < 0.001), IDH mutations (p < 0.001), and 1p/19q co-deletions (p = 0.01). ITSS grades 2–3 were linked to glioblastomas (p < 0.001), necrosis (p < 0.001), microvascular proliferation (p < 0.001), and CDKN2A/B homozygous deletions (p = 0.02). Models combining ITSS with rCBV and volume showed AUC of 0.94 and 0.96 (p < 0.001), outperforming univariate models. Conclusions: Semiquantitative ITSS grading correlates with key histopathologic and molecular glioma features. Combined with perfusion and volumetric parameters, ITSS enhance non-invasive glioma grading, in alignment with WHO CNS5. Full article

This study develops a landslide susceptibility zoning map for the Zipaquirá–Pacho road corridor in Cundinamarca, an area prone to frequent landslides. Two statistical methods—Weight of Evidence (WoE) and Information Value (IV)—were used alongside various causal factors to generate the map using GIS software (ArcGIS Pro 3.5.0 software.). A landslide inventory with 101 points was compiled through fieldwork and Google Earth image analysis. Of these, 70% were used to build the models, while the remaining 30% were reserved for validation, ensuring spatial representativeness. The resulting susceptibility maps classified the area into five categories: “very high”, “high”, “moderate”, “low”, and “very low.” For WoE, 19.62% of the area was classified as “very high” and 19.71% as “high”, while for IV, the respective values were 17.57% and 26.55%. Notably, 88% of the identified landslides occurred in “high” and “very high” zones. Model validation using the AUC (Area Under Curve) metric yielded an efficiency of 81%, confirming the reliability of both methods for landslide prediction. The study’s findings are essential for supporting mitigation strategies and serve as valuable input for local authorities and stakeholders involved in risk management and infrastructure planning. Full article

Introduction: Hyperbilirubinemia in newborns can lead to kernicterus, a severe form of neonatal encephalopathy caused by bilirubin toxicity. Despite timely interventions such as exchange transfusion, kernicterus can still develop, especially in high-risk infants. MRI is crucial for detecting early and evolving signs of bilirubin-induced brain damage. Case Report: We report a term newborn who developed severe hyperbilirubinemia and kernicterus despite receiving exchange transfusion. The infant presented on day 3 of life with jaundice, hypotonia, and feeding difficulties and had a bilirubin level of 51 mg/dL. After exchange transfusion, bilirubin levels normalized, but neurotoxicity persisted. Initial MRI at one month showed mild T1 hyperintensity in the hippocampi with no changes in the basal ganglia. At two months, T1 hyperintensities in the hippocampi partially resolved. By six months, MRI revealed T2 hyperintensities in the globus pallidus and hippocampal atrophy, consistent with kernicterus. Magnetic resonance spectroscopy (MRS) showed reduced N-acetylaspartate (NAA) levels, indicating neuronal loss. Discussion: MRI is essential in monitoring bilirubin-induced brain injury. In this case, early MRI findings showed mild hippocampal T1 hyperintensity, which resolved partially. At six months, T2 hyperintensities in the globus pallidus confirmed chronic bilirubin encephalopathy. MRS demonstrated a reduction in N-acetylaspartate, indicative of neuronal loss. Susceptibility-Weighted Imaging (SWI) showed no abnormalities, likely due to the myelination process in neonates. Conclusions: This case highlights the importance of repeated MRI in detecting bilirubin-induced brain damage. Early neuroimaging enables timely interventions and improves long-term neurodevelopmental outcomes in infants with severe hyperbilirubinemia. Full article

Eye diseases (EDs), including glaucoma, diabetic retinopathy, and cataracts, are major contributors to vision loss and reduced quality of life worldwide. These conditions not only affect millions of individuals but also impose a significant burden on global healthcare systems. As the population ages and lifestyle changes increase the prevalence of conditions like diabetes, the incidence of EDs is expected to rise, further straining diagnostic and treatment resources. Timely and accurate diagnosis is critical for effective management and prevention of vision loss, as early intervention can significantly slow disease progression and improve patient outcomes. However, traditional diagnostic methods rely heavily on manual analysis of fundus imaging, which is labor-intensive, time-consuming, and subject to human error. This underscores the urgent need for automated, efficient, and accurate diagnostic systems that can handle the growing demand while maintaining high diagnostic standards. Current approaches, while advancing, still face challenges such as inefficiency, susceptibility to errors, and limited ability to detect subtle retinal asymmetries, which are critical early indicators of disease. Effective solutions must address these issues while ensuring high accuracy, interpretability, and scalability. This research introduces a 2D single-channel convolutional neural network (CNN) based on ResNet101-V2 architecture. The model integrates gradient-weighted class activation mapping (Grad-CAM) to highlight retinal asymmetries linked to EDs, thereby enhancing interpretability and detection precision. Evaluated on retinal Optical Coherence Tomography (OCT) datasets for multiclass classification tasks, the model demonstrated exceptional performance, achieving accuracy rates of 99.90% for four-class tasks and 99.27% for eight-class tasks. By leveraging patterns of retinal symmetry and asymmetry, the proposed model improves early detection and simplifies the diagnostic workflow, offering a promising advancement in the field of automated eye disease diagnosis. Full article

Susceptibility-weighted imaging (SWI) is a magnetic resonance imaging (MRI) sequence sensitive to substances that alter the local magnetic field, such as calcium and iron, allowing phase information to distinguish between them. SWI is a 3D gradient–echo sequence with high spatial resolution that leverages both phase and magnitude effects. The interaction of paramagnetic (such as hemosiderin and deoxyhemoglobin), diamagnetic (including calcifications and minerals), and ferromagnetic substances with the local magnetic field distorts it, leading to signal changes. Neurodegenerative diseases are typically characterized by the progressive loss of neurons and their supporting cells within the neurovascular unit. This cellular decline is associated with a corresponding deterioration of both cognitive and motor abilities. Many neurodegenerative disorders are associated with increased iron accumulation or microhemorrhages in various brain regions, making SWI a valuable diagnostic tool in clinical practice. Suggestive SWI findings are known in Parkinson’s disease, Lewy body dementia, atypical parkinsonian syndromes, multiple sclerosis, cerebral amyloid angiopathy, amyotrophic lateral sclerosis, hereditary ataxias, Huntington’s disease, neurodegeneration with brain iron accumulation, and chronic traumatic encephalopathy. This review will assist radiologists in understanding the technical framework of SWI sequences for a correct interpretation of currently established MRI findings and for its potential future clinical applications. Full article

A fundamental limitation of fMRI based on the BOLD effect is its limited spatial specificity. This is because the BOLD signal reflects neurovascular coupling, leading to macrovascular changes that are not strictly limited to areas of increased neural activity. However, neuronal activation also induces microstructural changes within the brain parenchyma by modifying the diffusion of extracellular biological water. Therefore, diffusion-weighted imaging (DWI) has been applied in fMRI to overcome BOLD limits and better explain the mechanisms of functional activation, but the results obtained so far are not clear. This is because a DWI signal depends on many experimental variables: instrumental, physiological, and microstructural. Here, we hypothesize that the γ parameter of the fractional diffusion representation could be of particular interest for DW-fMRI applications, due to its proven dependence on local magnetic susceptibility and diffusion multi-compartmentalization. BOLD fMRI and DW-fMRI experiments were performed at 3T using an exemplar application to task-based activation of the human visual cortex. The results, corroborated by simulation, highlight that γ provides complementary information to conventional diffusion fMRI and γ can quantify cellular morphology changes and neurovascular regulation during neuronal activation with higher sensitivity and specificity than conventional BOLD fMRI and DW-fMRI. Full article

Background: The brain’s vascular system has recently been shown to provide an important efflux pathway for cerebral waste clearance (CWC). However, little is known about the influence of aging or diabetes on the CWC. The aim of the current study is to investigate the vasculature contribution to CWC under aging and diabetic conditions. Methods: Male Wistar rats under aging and diabetic conditions were evaluated using dynamic intra-cisterna superparamagnetic iron oxide-enhanced susceptibility-weighted imaging (SPIO-SWI). Theoretical analysis of the expected signal intensity using SPIO-SWI was compared with the corresponding dynamic in vivo images. Quantitative susceptibility mapping (QSM) was used to evaluate the iron-based tracer concentration in the venous system. Results: Our data demonstrated that the theoretical analysis predicted the dynamic changes in the signal intensity after SPIO infusion. The distinct hyperintense signals due to the lower concentration of the SPIO over time in cerebrospinal fluid (CSF) and meningeal lymphatic (ML) vessels likely represented the CWC through various efflux pathways, including cerebral vascular and ML vessels. The QSM analysis further revealed reduced CWC from the vasculature in both the aged and diabetic groups compared to the younger group. Conclusions: Our results demonstrated that SPIO-SWI can quantitatively evaluate the CWC efflux contributions from cerebral vascular vessels under aging or diabetic conditions. Full article