This pilot study aimed to investigate the feasibility, safety, and preliminary efficacy of a rehabilitation program utilizing Syrebo™ SY-HR03E soft robotic glove with pneumatic actuators for hand function recovery in patients with early subacute stroke. A single-blind pilot randomized controlled trial was conducted in an inpatient clinical setting. The intervention was administered 5 days per week for 4 weeks (20 session’s total). During this period, the robotic therapy (RT) group (n = 10) received 30-min daily sessions with the soft robotic glove, while the conventional therapy (CT) group (n = 10) received 30-min daily sessions of conventional hand therapy. Both groups additionally participated in an identical 1-h daily conventional rehabilitation program excluding hand training. The outcome measures included the Fugl–Meyer Assessment for the hand (FM-Hand) and upper limb (FM-UL), the modified Barthel index (MBI), and the Brunnstrom stages (BS). Intervention effects were analyzed using two-way repeated-measures ANOVA. All participants completed the intervention without dropouts or adverse events, confirming the feasibility and safety of the protocol. Baseline motor impairment was comparable between groups (FM-Hand: 2.20±1.93 vs. 2.50±2.64; FM-UL: 19.60±14.76 vs. 16.70±14.50). A significant Group × Time interaction was found for the FM-Hand [F(1,18) = 4.743, p < 0.05] and the MBI [F(1,18) = 7.728, p < 0.05]. Simple effects analysis revealed that both groups improved significantly in FM-Hand, while only the RT group showed a significant improvement in MBI after interventions. For the FM-UL, a significant main effect of Time [F(1,18) =24.931, p < 0.001] in both groups was observed without an interaction. In the BS, a significant main effect of Time was found for both hand and upper limb, with only the RT group showing significant within-group improvement for the hand. SyreboTMSY-HR03E soft robotic glove is a feasible and safe tool for hand rehabilitation in early subacute stroke. The results provide preliminary signals for improving hand function and translating gains into activities of daily living compared to conventional therapy alone. However, because both groups were in the early subacute stage, the observed gains likely reflect a combination of intervention effects and spontaneous recovery. These findings justify further investigation through larger-scale, definitive trials. Trial Registration: ChiCTR2000034614||http://www.chictr.org.cn/ with the Clinical Trial Registry (2020/07/12).
Female fertility is declining globally, which has become a serious challenge for social development. Currently, the main methods of female fertility cryopreservation include cryopreservation of immature or mature oocytes, embryos, and ovarian tissue. Among them, ovarian tissue vitrification is regarded as a promising method to preserve female fertility, especially for children or adolescent women. Therefore, ovarian tissue vitrification has attracted more and more attention in recent years. Through the efforts of researchers, the technology has evolved from simple animal experiments to applications in humans. There have been cases of revived transplantation of cryopreserved ovarian tissue with successful pregnancies. However, cryo-injury, oxidative stress and other related injuries during vitrification can seriously reduce the quality of ovarian grafts and affect the therapeutic efficacy of transplantation, which are still major challenges for ovarian tissue vitrification. The purpose of this review is to summarize the challenges in the current protocols, propose relevant solutions, and provide guidance for ovarian tissue vitrification to help improve female fertility cryopreservation.
Bisphenol A (BPA) is an environmental toxicant known to induce nephrotoxicity primarily through oxidative stress–mediated inflammation and apoptosis. Mesenchymal stem cell-derived exosomes have recently emerged as promising cell-free therapeutic agents with antioxidant and anti-apoptotic properties. This study aimed to investigate the protective effects of bone marrow mesenchymal stromal cell-derived exosomes against BPA-induced nephrotoxicity in human proximal tubular epithelial (HK-2) cells, with particular emphasis on the potential involvement of the Nrf2/Keap1 signaling pathway. HK-2 cells were allocated into four groups: control, BPA, exosome (EXO), and BPA + EXO. Cell viability was assessed using the MTT assay. Intracellular reactive oxygen species (ROS), lipid peroxidation, antioxidant parameters, total antioxidant capacity (TAC), glutathione (GSH) levels, and inflammatory cytokines (IL-6 and IL-10) were evaluated using standard biochemical assays. Protein expression levels of Nrf2, Keap1, Bax, and Bcl-2 were determined by Western blot analysis. BPA exposure significantly reduced cell viability and antioxidant capacity while increasing ROS production, lipid peroxidation, pro-inflammatory cytokine release, and apoptotic signaling. Exosome treatment markedly attenuated oxidative stress, restored antioxidant defenses, improved inflammatory cytokine balance, and reversed BPA-induced alterations in Nrf2/Keap1 signaling and apoptosis-related protein expression. Bone marrow mesenchymal stromal cell-derived exosomes exert protective effects against BPA-induced cytotoxicity in HK-2 cells by alleviating oxidative stress, modulating the Nrf2/Keap1 signaling pathway, and suppressing apoptosis. Although further mechanistic studies are required to establish causality, these findings highlight the therapeutic potential of exosomes as a cell-free strategy for mitigating environmental toxin-associated renal injury.
Ionizing radiation poses a significant threat to human health and necessitates the development of effective radioprotective agents. In this study, the radioprotective efficacy of ginsenoside Rh2-loaded casein nanoparticles (Rh2-CNPs) was synthesized and evaluated. The Rh2-CNPs were prepared using a modified co-precipitation method. Physicochemical characterization revealed nanoparticles with an average hydrodynamic diameter of 180 ± 15 nm, a zeta potential of -28 ± 2 mV, an encapsulation efficiency of 75.3 ± 8.5
Stem cell sheets have unique cell-to-cell connections, the ability to enhance and preserve the extracellular matrix (ECM) and cell–ECM interactions, and a scaffold-free nature, which makes them ideal for regenerative medicine. This study aims to explore the potential of stem cell sheets for osteochondral construct fabrication. Mesenchymal stem cells (MSCs) were isolated from rat bone marrow and differentiated into osteogenic and chondrogenic sheets. These differentiated cell sheets were subsequently stacked using a collagen and hyaluronic acid hydrogel between the layers to create multilayered constructs. Histological analysis demonstrated notable tissue formation of a multilayered structure with distinct osteogenic and chondrogenic layers both with and without the hydrogel constructs. Scanning electron microscopy (SEM) confirmed the integrity and content of the sheets. qRT-PCR and immunocytochemistry analysis demonstrated the presence of key related bone and cartilage markers within the differentiated sheets. In conclusion, our data emphasize that stem cell sheet engineering (CSE) holds significant promise for developing 3D, scaffold-free constructs for osteochondral regeneration. The ability to create relatively distinct osteogenic and chondrogenic layers in constructs has an extreme impact on the field of regenerative medicine, leading to a new therapeutic approach and hope for the future of osteochondral defects treatment.
A flexible and promising method for creating polymeric nanofibrous composites for BTE (bone tissue engineering) is electrospinning. The optimal composites should resemble bone’s natural ECM (extracellular matrix), possess suitable surface chemistry, exhibit strong mechanical strength, and demonstrate bioactivity and biocompatibility. Selecting the appropriate material for composite fabrication is a critical initial step in developing a BTE (tissue engineered) product. BTE is a multidisciplinary area that applies engineering concepts to bone-related biological processes. Key considerations in this field include composites, cells, growth factors, and their interfaces in the microenvironment. This review article examines the recent developments in BTE using biomimetic electrospun polymeric biomaterials. It provides an overview of Natural polymer, synthetic polymer, bone structure, biocomposite discusses the ideal characteristics of bone scaffolds, and offers a basic introduction to BTE. The fabrication of bone composites is explored in detail, including various materials and techniques used. Additionally covered the characteristics, benefits, and uses of electrospinning in BTE. The application of electrospun nanofibrous composites composed of natural and synthetic polymers for BTE is thoroughly reviewed and discussed, as well as the biomineralization processes involved. Finally, the study concludes with insights into biomimetic composites for BTE based on electrospun nanofibers.
To investigate the efficacy of transcatheter arterial chemoembolization (TACE) combined with targeted immunotherapy in the treatment of advanced liver cancer. A total of 115 patients with intermediate-advanced liver cancer treated in our hospital from June 2021 to June 2024 were retrospectively selected and divided into two groups according to the actual treatment regimen received. The grouping was a non-randomized retrospective design. The conventional group (n = 57) received TACE alone, while the combination group (n = 58) received TACE combined with targeted immunotherapy. Clinical efficacy, tumor markers [alpha-fetoprotein (AFP), carcinoembryonic antigen (CEA), carbohydrate antigen 125 (CA125), carbohydrate antigen 199 (CA199), protein induced by vitamin K absence-II (PIVKA-II)], immune function [immunoglobulin A (IgA), immunoglobulin G (IgG), immunoglobulin M (IgM)], and liver function indicators [serum albumin (ALB), glutamic-pyruvic transaminase (GPT), total bilirubin (TBIL)] were compared between the two groups. Kaplan–Meier survival analysis and the Cox proportional hazards regression model were used to evaluate survival rate, the incidence of adverse events was also compared. To control for potential confounding factors, 1:1 propensity score matching (PSM) was performed as a sensitivity analysis, subgroup analyses of efficacy and safety for different targeted drugs within the combination group were also carried out. Immunohistochemistry was used to detect the expression of Ki-67, CD8 ⁺ T, and PD-L1 in two groups before treatment. Compared with the conventional group, the proportion of objective response rate (ORR), disease control rate (DCR) patients in the combined group was significantly higher (P < 0.05). After treatment, the levels of tumor markers in both groups were significantly reduced, which were lower in the combined group than the conventional group (P < 0.001). After treatment, the immune function in both groups were improved, with the combined group showing superior efficacy compared to the conventional group (P < 0.001). After treatment, the liver function in both groups were improved. The ALB level in the combined group was significantly higher than that in the conventional group, while the GPT and TBIL levels were significantly lower than those in the conventional group (P < 0.001). The one-year cumulative survival rate of patients in the combined group was higher than that of the conventional group (P < 0.05). The overall incidence of grade ≥ 3 adverse events did not differ significantly between the two groups (P > 0.05). After PSM, the DCR and 1-year cumulative survival rate in the combination group remained significantly superior to those in the conventional group (both P < 0.05). Combination therapy was an independent protective factor for survival (HR = 0.520, 95
Falls are a frequent event among older adults and individuals with central neurological conditions. Gaze behaviour (i.e. how individuals use visual input to guide movement) plays a critical role in motor control while walking and likely contributes to fall prevention. However, existing research in this area remains fragmented. This scoping review will aim to identify how gaze behaviour is measured and utilized during gait rehabilitation among older adults and individuals with central neurological conditions. Findings will inform standardized assessment approaches and the development of interventions targeting gaze behaviour to improve walking and reduce fall risk. This review will follow the Joanna Briggs Institute scoping review guidelines and is registered with the Open Science Framework. Studies will be included if they involve older adults (≥ 65 years) or individuals with central neurological conditions and measure gaze behaviour during walking. A comprehensive search strategy was developed in collaboration with a University Health Network Librarian and refined iteratively by the research team. Eight databases (Medline, Embase, CENTRAL, CDSR, Emcare Nursing, APA PsycINFO, Web of Science Core Collection, and CINAHL) were searched from inception to August 2025. Title and abstract screening will be followed by a full-text review to identify articles that meet the inclusion criteria. Data related to study/participant characteristics, methods for eye tracking, intervention, equipment, outcome measures, and key findings will be extracted. Methodological quality of studies will be assessed using the appropriate tool based on study design. Results will be reported according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses extension for Scoping Reviews checklist.
Quantitative kinematic tracking of in vivo skeletal structures efficiently and with high temporal resolution is important to a wide range of questions in biomechanical research. As such, software tools are needed for efficient, semi-automated registration of bones imaged with either 3DCT or 4DCT. This study presents Hierarchical 3D Registration (3DH), an open-source approach to skeletal tracking of three-dimensional image volume sets. The 3DH approach and software are presented using sequential 3DCT datasets of participants performing various thumb tasks and dynamic 4DCT datasets of simulated flexion–extension in cadaveric wrists collected in previous studies. The agreement in computed arthrokinematics with previously calculated values using independent approaches for 3DCT and 4DCT data was assessed with Bland–Altman analyses. Using 3DH, all target bones were successfully tracked for both the 3DCT and 4DCT sets. For 3DCT data of the radius, first metacarpal and trapezium, the mean bias for the helical angle was −0.06 degrees with 95
Single-use gastrointestinal endoscopes eliminate the need for post-procedure reprocessing and have become an area of interest in endoscopic device development and quality management. However, clinical data on their use in routine gastrointestinal endoscopy remain limited. This prospective, two-center, single-arm study evaluated the feasibility and short-term safety of single-use gastroscopy performed with ENDOANGEL, an artificial intelligence (AI)-based real-time quality monitoring system, and single-use colonoscopy in routine gastrointestinal endoscopy. Formal AI-based quality scoring was applied to gastroscopy, whereas ENDOANGEL-related colonoscopy indicators were recorded descriptively. A total of 120 participants were enrolled, including 60 who underwent gastroscopy and 60 who underwent colonoscopy. Imaging of predefined anatomical sites was successful in all participants in both groups (60/60 in each group; 95
Conventional identification of stable HFpEF still depends on resource-intensive clinical assessment, whereas non-invasive acoustic analysis of sustained vowels may provide a complementary and accessible classification signal. The objective of this study was to develop and validate a classification model for stable heart failure with preserved ejection fraction (HFpEF) versus healthy controls using acoustic features extracted from the sustained vowel /ɑː/. In this retrospective case–control study, voice recordings were obtained from a primary cohort of 341 participants and an independent external validation cohort of 172 participants. The present study extracted 384 features using the Interspeech 2009 feature set and compared five deep learning models. The top-performing deep learning architecture, the multilayer perceptron (MLP), was further assessed through tenfold cross-validation, independent external validation, and feature analysis using SHapley Additive exPlanations and Local Interpretable Model-Agnostic Explanations. Additional analyses benchmarked the MLP against classical machine-learning models and examined uncertainty, calibration, prevalence sensitivity, and demographic confounding. For the binary classification of stable HFpEF versus healthy controls, the MLP classifier demonstrated the strongest performance in the five-model deep learning comparison, achieving a mean tenfold cross-validation accuracy of 0.8593 and an area under the curve (AUC) of 0.9130. On the independent external validation cohort, the primary MLP achieved an AUC of 0.836. In a broader benchmark, acoustic-feature models clearly outperformed age/sex-only models, and MLP remained competitive against strong classical tabular baselines. A regularized MLP sensitivity model achieved an external AUC of 0.8638 (95
High-density grid electrodes can be configured to spatially scale the neurostimulation fields and enable the control of depth, intensity, and stimulation vectoring. In this study, we computationally analyze a high-density cluster electrode (HDCE) technology to measure its scalability and its effects on human tissue. The models are developed in Ansys using the Maxwell 3D Design modeler. More than 24 HDCE electrode configurations were analyzed and compared to standard 8 mm Ag/AgCl electrodes. It was found that the HDCE electrode configurations can be selected to emulate the electric field properties to within 1
In silico models for simulating bone growth based on mechanical or non-mechanical epigenetic factors are widely used. In this study, a well-known mechanobiological model, which states that octahedral shear stress accelerates longitudinal bone growth and hydrostatic stress retards it, is applied to a finite element model of the femur of an 8-year-old boy. Proximal and distal epiphyseal plates as well as the growth plate of the greater trochanter, cartilaginous growth at the femoral isthmus, and appositional bone growth are included in the model. Furthermore, changes in the density of the cancellous bone in the metaphyses are modeled based on Wolff's law using compressive stresses as the mechanical stimulus. Muscle forces during a dynamic gait cycle were determined for nine discrete loading cases by optimizing to minimize bending stress. The highest stresses in the femoral shaft were determined as medial compressive stresses with a maximum of −33.2 MPa. Highest internal axial load in the shaft was 985 N during loading response. The simulated bone growth resulted in an increase in femur length of 23.7 mm and a decrease in femoral neck angle by −1.4°, anteversion angle by −2.4°, and lateral distal femur angle by −1.6° per year. Growth of the apophysis of the greater trochanter resulted in an unchanged articulo-trochanteric distance. The bone remodeling led to an increase in bone density, particularly in the medial proximal metaphysis. The consideration of different growth mechanisms allowed a comprehensive simulation of femoral growth with high agreement with anthropometric data. Possible applications are the simulation of the correction of deformities.
Lung cancer ranks among the most lethal malignancies globally, and its traditional diagnosis suffers from strong subjectivity, high misdiagnosis rates and uneven medical resources. To overcome the poor feature alignment caused by simply concatenating CT images and clinical text, this paper proposes a lung cancer multimodal auxiliary diagnosis model based on entropy weight decision fusion. This retrospective cohort study enrolled 5847 participants from 2020 to 2025, comprising 1823 lung cancer cases, 2253 normal controls and 1771 pulmonary nodule controls. All CT images and corresponding reports were analyzed, and three datasets were established via random sampling from the original dataset. The study incorporated Vision Transformer (ViT) and Bidirectional Encoder Representations from Transformers (BERT) as feature extractors for images and text, respectively, to extract high-dimensional semantic features from lung CT images and CT Imaging Report. Secondly, independent classifiers based on Multi-Layer Perceptron (MLP) were established to convert the embedding vectors of different modalities into predicted probability distributions (Logits). Finally, the entropy weight method was employed to adaptively fuse the decision results of images and text. The model performance evaluation indicators include area under the receiver operating characteristic curve (AUC), accuracy, precision, recall, and F1-score. The proposed method in this study can fully leverage the complementary information from CT images and imaging text multimodality. On the clinical dataset, it achieved an accuracy of 0.9375, a precision of 0.9324, a recall of 0.9322, and an F1-score of 0.9322, significantly improving diagnostic performance. This study validates that, on a real-world lung cancer dataset, multimodal data decision fusion outperforms unimodal models and common fusion methods in terms of diagnostic accuracy, precision and recall. It provides a potential reference for the early auxiliary diagnosis of pulmonary nodules and lung cancer, and lays a foundation for subsequent clinical applications.
Bone regeneration assisted by synthetic bone substitutes largely depends on the integration of the vascular, neural, and lymphatic systems in the bone. Bone marrow mesenchymal stem cells (BMSCs) are the key cells for this process. However, their role in regulating the integration has not been fully characterized. Human BMSCs (hBMSCs) were treated with osteogenic induction and collected from 0 to 504 h for bulk RNA sequencing (RNA-Seq). Differentially expressed genes (DEGs) were identified and Gene Ontology (GO), Kyoto Encyclopedia of Genes and Genomes (KEGG) and Time-Series Transcriptomic Trend Analysis were used to comprehensively analyze the possible pathways and functions associated with these DEGs. Weighted Gene Co-expression Network Analysis (WGCNA) was constructed to identify the modules and hub genes of the process. Quantitative real-time polymerase chain reaction (qRT-PCR) and enzyme-linked immunosorbent assay (ELISA) were performed to validate the expression of key genes identified by RNA-Seq. Time-series analysis of the hBMSCs transcriptome suggested a dynamic expression trajectory during osteogenic differentiation, which was characterized by four functional patterns: the initial adaptation stage (1–24 h), the proliferation activation stage (24–72 h), the differentiation regulation stage (72–336 h) and the remodeling stability stage (336–504 h). Moreover, 72 h was suggested as a potential key time point in the osteogenic–vascular–neural–lymphatic coupling process based on transcriptomic analysis, with typical activation of BMP, vascular endothelial growth factor (VEGF) and PPAR signaling pathways. Four modules and closely related hub genes such as growth differentiation factor 5 (GDF5), matrix Gla protein (MGP) and pregnancy-associated plasma protein A2 (PAPPA2), whose expressions were validated by qRT-PCR and ELISA were also identified and highlighted. Our study revealed the temporal trends of angiogenesis, lymphangiogenesis, and neurogenesis during BMSCs osteogenic differentiation, which not only supplemented the transcriptional regulation in bone regeneration, but also provided a theoretical basis for the design of synthetic bone substitutes.
Ceramic materials are widely used in bone tissue engineering applications due to their chemical similarity to human bone, along with their inherent biocompatibility. Bone scaffolds fabricated with ceramics thus serve as a framework for supporting new bone formation through cell attachment, proliferation, and differentiation, and subsequently integrate with the host tissue. However, the cells are unable to attach directly to the implanted ceramic surfaces. Instead, they interact with the dynamic layer of proteins adsorbed on the material surface, which mediates cell adhesion and proliferation via specific integrin-ligand signalling. Therefore, rather than a passive process, protein adsorption can be utilized as a design criterion to develop advanced ceramic scaffolds. Despite extensive efforts in surface modification, the mechanistic relationship between ceramic surface properties, protein adsorption behaviour, and subsequent osteogenic signalling remains fragmented across the literature. This review emphasizes the central and often underexplored role of protein adsorption in mediating the initial cell–ceramic interactions critical for bone tissue regeneration. In contrast to previous pieces of literature, this paper critically examines the studies in the past decade, with special focus on the last five years, on how the protein–ceramic interaction can be manipulated to improve its biocompatibility, which adds to the novelty of this review. By critically evaluating the in vitro and in vivo studies, we propose that the protein adsorption on ceramic scaffolds can be treated as a controllable bio-instructive design parameter for next-generation osteoinductive ceramics. This review highlights the multifaceted nature of protein adsorption and its pivotal role in developing more biocompatible ceramic materials for bone tissue engineering.
Accurate segmentation of acute ischemic stroke (AIS) lesions on neuroimaging is essential for diagnosis, treatment decision-making, and prognostication. Manual methods are limited by time and variability. Machine learning (ML), especially deep learning, has emerged as a powerful tool for automated lesion segmentation, yet a systematic synthesis of model performance, methodological rigor, and clinical applicability remains lacking. To systematically review and quantitatively evaluate the performance of ML-based segmentation models for AIS using a meta-analytic approach, and to identify factors associated with model accuracy and robustness across imaging modalities, architectures, and datasets. We conducted a systematic review and meta-analysis in accordance with PRISMA 2020 guidelines. Comprehensive searches were performed in PubMed, Scopus, and Web of Science databases through March 2025. Eligible studies included those reporting on machine learning (ML)-based segmentation of acute ischemic stroke (AIS) lesions on CT or MRI and providing quantitative performance metrics (e.g., Dice, sensitivity, specificity, AUC). Data were systematically extracted on study design, ML architecture, imaging modality, dataset size and composition, and segmentation performance. Random-effects meta-analyses were conducted using inverse-variance weighting, with logit transformation applied to bounded metrics to stabilize variance. Between-study heterogeneity was assessed using the I2 statistic and Cochran’s Q-test. Meta-regression analyses explored the influence of covariates such as lesion volume, sample size, and stroke severity (mRS), while subgroup analyses examined performance variations by imaging modality and model type. Visualizations included forest plots, funnel plots, bubble plots, and correlation matrices, generated in Python using standardized meta-analysis and statistical libraries. Out of 4755 screened records, 101 studies met the inclusion criteria. Deep learning approaches, especially U-Net variants, dominated the field (78
Mechanical forces significantly influence the initiation, progression, and remodeling of cardiovascular disease. In this study, CT-based patient-specific computational models were constructed to investigate the effects of the successful septal myectomy on myocardial stress distribution in the left ventricle. CT imaging data from five patients, collected both before and after successful septal myectomy, were used to create 10 patient-specific finite element models. Myocardial stress and strain of all integral nodes in the myocardium of the left ventricle were extracted from the simulation results. All the nodes in the left ventricular wall were divided into apex, midventricular, and basal region groups according to their position in the left ventricle. It was found that the mean stress decreased by 27.7
Adolescent Idiopathic Scoliosis (AIS) is a common spinal deformity arising during adolescence, a critical period for brain development. Although most AIS analyses predominantly focus on spinal or behavioral aspects, it remains unclear what brain functional alteration is associated with AIS. This gap in knowledge impedes a holistic understanding of the condition’s pathophysiology. In this study, functional near-infrared spectroscopy (fNIRS) was used to explore the resting-state brain network in AIS. The study recruited 25 AIS patients and 25 age-matched healthy controls and measured their brain activities during resting-state lying using fNIRS. Brain functional connectivity and network topology were evaluated for the two groups and their correlations with demographic and pathological variables were examined. The functional connectivity in the patients, particularly single-curve patients, decreased and was sparser in the parietal and prefrontal regions in comparison to the healthy controls. The regional nodal metrics in the patients were significantly altered, with smaller nodal degrees and efficiency observed in certain nodes. Among the affected regions, the dorsolateral prefrontal cortex emerged repeatedly as a hub showing altered connectivity in patients, particularly in relation to parietal and sensorimotor regions. In addition, different correlations between brain network metrics and demographic as well as pathological parameters were identified within patients. For instance, a larger primary Cobb angle was associated with poorer small-world properties, suggesting greater structural deformity may be linked to less efficient functional network organization. Furthermore, a support vector machine-based classification model using functional connectivity achieved an average accuracy of 81.0
Magnetic resonance imaging (MRI) is widely regarded as the most reliable non-invasive imaging modality for detecting neurological disorders. However, manual interpretation of MRI scans is often time-consuming and prone to inter-observer variability, which can lead to inconsistencies in diagnosis. The global burden of neurological disorders—including Alzheimer’s disease, brain tumors, Parkinson’s disease, multiple sclerosis, and schizophrenia—continues to increase, creating an urgent demand for accurate, scalable, and automated diagnostic solutions. In recent years, deep learning (DL) has emerged as a powerful paradigm for medical image analysis, enabling automated feature extraction and improved diagnostic performance in neuroimaging applications. This survey provides a comprehensive analysis of deep learning approaches for MRI-based detection of neurological disorders. A systematic review of 47 research articles published between 2019 and 2025 is conducted, covering over 40 deep learning architectures evaluated on 34 publicly available and clinical datasets. The study categorizes and critically examines convolutional neural networks (CNNs), Vision Transformers (ViTs), hybrid CNN–Transformer models, and other specialized deep learning frameworks developed for neurological disease detection and classification. Comparative analyses are presented across different neurological conditions, highlighting model performance, dataset characteristics, evaluation protocols, and computational requirements. Furthermore, the survey identifies emerging architectural trends and evaluates the relative strengths and limitations of existing approaches with respect to generalization, interpretability, and clinical applicability. Key research gaps are highlighted, including limited cross-institutional validation, dataset heterogeneity, insufficient explainability, and challenges in real-world clinical deployment. Finally, the paper outlines promising research directions such as multimodal learning, self-supervised representation learning, and robust cross-domain generalization to enhance the reliability and clinical translation of MRI-based deep learning systems.
