搜索结果：image

Since the various contrast-weighted MR images of a given anatomy contain redundant information, one contrast can be used to guide the reconstruction of another undersampled contrast acquired subsequently in the same session. To solve this reconstruction problem leveraging multi-contrast side information, several end-to-end learning-based guided reconstruction methods have been proposed. However, a key challenge is the requirement for large paired training datasets comprising raw k-space data and aligned reference images. We propose a modular plug-and-play approach, which requires no k-space training data and relies solely on partially paired image-domain datasets. In this approach, a content/style model of two-contrast MR data is first learned from a purely image-domain dataset and subsequently applied as a plug-and-play operator in iterative reconstruction. The disentanglement of content and style allows explicit representation of contrast-independent and contrast-specific factors. Consequently, incorporating prior information into the reconstruction reduces to a simple replacement operation on the aliased content of the estimated image using high-quality content derived from the reference scan. Combining this so-called content consistency operation with an MR data consistency step, followed by a corrective procedure for the content estimate, yields an iterative scheme. We name this novel approach PnP-CoSMo. This approach, by design, offers cross-contrast generalizability and provides an explanatory framework based on the shared and non-shared generative factors underlying the two given contrasts. We explore various aspects of PnP-CoSMo, including interpretability and convergence, via simulations. Furthermore, its practicality is demonstrated on the public NYU fastMRI DICOM dataset, showing equivalent or superior quality and greater generalizability compared to end-to-end methods. On two in-house multi-coil datasets, PnP-CoSMo enabled up to 32.6% greater acceleration over non-guided plug-and-play reconstruction at given SSIM.

Cognitive reappraisal involves reinterpreting negative content to reduce its negative impact. Recent findings suggest that cognitive reappraisal can attenuate the link between emotions and eating-related measures. The level of regulatory engagement during the use of cognitive reappraisal can be assessed by physiological markers such as pupil dilation. The present study examined whether trial-by-trial within-person variations in pupil diameter while implementing reappraisal can predict subsequent desire to eat. Forty-three healthy females completed a computerized task combining a standard reappraisal task with a food-rating task. Participants viewed negative or neutral non-food-related images and were instructed to observe them passively or reappraise their content. Each image was followed by a picture of a food item, and participants rated their desire to eat the depicted food. Observing negative images led to an immediate reduction in the desire to eat compared to viewing emotionally neutral images. Importantly, applying reappraisal reduced the impact of negative emotions on the desire to eat. Moreover, greater pupil dilation during reappraisal, but not during passive viewing of negative or neutral images, predicted higher subsequent desire to eat. These findings suggest that regulatory engagement plays a key role in shaping the relationship between emotions and the desire to eat. The study offers insights relevant to understanding disorders involving both dysregulated eating and difficulties applying cognitive reappraisal.

Deep learning models often struggle with class imbalance and low-resolution medical images, where critical spatial details and minority-class features are underrepresented. We introduce the Adaptive Distribution-aware Vision Transformer (AdaptiveViT), a novel hybrid CNN-Transformer architecture that unifies fine-grained local feature extraction with global contextual modelling. AdaptiveViT incorporates a distribution-aware modulation mechanism that adaptively adjusts feature emphasis according to the severity of class imbalance. In addition, a Distribution-aware Adaptive (DA) Loss incorporates the dataset imbalance ratio into an adaptive focusing scheme, enhancing minority-class sensitivity. Experiments on five skin lesion datasets with varying image resolutions and imbalance ratios (as high as 1:10 for melanoma versus non-melanoma) demonstrate that AdaptiveViT consistently outperforms state-of-the-art CNN, Transformer, and hybrid baselines in F1 and AUC, while maintaining stable convergence across imbalance levels. Validation on gastrointestinal endoscopy datasets further demonstrates AdaptiveViT's domain-agnostic generalisation beyond skin lesion data, which share similar imbalance characteristics. All experiments are conducted using patient-disjoint splits, with a threshold-free evaluation protocol to ensure fair, unbiased, and clinically reliable comparisons. Overall, AdaptiveViT establishes a hybrid framework for medical image classification under class imbalance and image-resolution variability. The code is available at https://github.com/mmu-dermatology-research/AdaptiveViT.

To evaluate the feasibility of using convolutional neural networks (CNNs) and vision transformers (ViTs) to predict renal tumor pathology intraoperatively based on gross appearance. Intraoperative images were retrospectively extracted from surgical recordings of patients undergoing partial nephrectomy between 2008-2024. Static frames obtained prior to arterial clamping were curated and linked with final pathology. A ResNet50-based CNN and the General Surgery Vision Transformer (GSViT) were trained to classify six tumor types: clear cell RCC (ccRCC), papillary RCC (pRCC), chromophobe RCC (chRCC), hybrid oncocytic tumors, oncocytoma, and angiomyolipoma (AML). Models were trained with transfer learning, evaluated on held-out test data, and assessed using accuracy, AUC-ROC, and confusion matrices. A total of 443 images from 118 patients (136 surgeries) were analyzed, including ccRCC (n=149), pRCC (n=97), chRCC (n=42), hybrid tumors (n=81), oncocytoma (n=43), and AML (n=31). In binary classification, the CNN achieved the highest AUCs for ccRCC (0.74), chRCC (0.70), hybrid tumors (0.73), and AML (0.70). Multi-class CNN performance was more variable, with notable AUCs for pRCC (0.70) and oncocytoma (0.71). The GSViT model underperformed across most categories, demonstrating prediction bias toward ccRCC. Attempts to unfreeze pretrained backbones led to rapid overfitting, underscoring dataset limitations. CNN-based models demonstrate moderate ability to classify renal tumor pathology intraoperatively from gross appearance, providing proof of concept for AI-assisted surgical decision-making. Larger datasets and external validation are needed before clinical application.

Multi-label fundus disease classification aims to assign multiple ocular disease labels to bilateral fundus images, enabling automated screening and clinical decision support. In contrast to most existing computer-aided systems that process each eye independently and rely on shallow feature fusion, we explicitly model binocular structure and adaptively combine information from both eyes to handle complex, real-world fundus scenes. To this end, we introduce DualCrossAttnNet, a multi-label fundus disease classification network specifically designed for binocular analysis. In particular, we obtain high-resolution bilateral representations using an EfficientNet-B2 backbone and a cross-attention module that jointly reasons about spatial and channel-wise interactions between the left and right eyes. We further employ a gated fusion mechanism to adaptively weight the contributions of each eye, and an SE attention module to recalibrate channel responses before global aggregation. Coupled with a fundus-oriented preprocessing pipeline and a GeM-based classifier, the proposed framework can accurately predict multiple co-occurring ocular diseases from complex clinical fundus images. On the public ODIR-2019 dataset, DualCrossAttnNet substantially improves multi-label classification performance, achieving 88.20% accuracy, 90.98% F1, and 98.49% AUC, with a composite score of 92.73%. These results surpass recent CNN, GNN, and Transformer-based baselines by up to 20.37 percentage points in composite score, demonstrating that DualCrossAttnNet is an effective and scalable solution for intelligent fundus disease diagnosis.

Image-guided vacuum-assisted biopsy (VAB) is the standard of care for sampling suspicious breast lesions and depends on single-use, vendor-proprietary consumables produced by a concentrated set of manufacturers. The January 2026 Hologic Brevera 9-gauge needle recall and the subsequent FDA shortage designation exposed the fragility of this supply chain. To examine the three principal strategies available to breast imaging programs for mitigating consumable supply disruptions (strategic inventory stockpiling, vendor diversification, and protocol and procurement standardization), and to describe how standardization functions as the connective infrastructure enabling the other two at network scale. Stockpiling provides an immediate operational buffer; vendor diversification eliminates structural single-source dependency; standardization allows both to operate consistently across distributed sites. No single strategy is sufficient alone. Enterprise-scale measures such as cross-site inventory redistribution presuppose centralized governance and resources, and continuity planning should hold community sites to the same standard as the primary campus.

Many intracellular pathogens manipulate host cell calcium to facilitate their survival and replication. Live-cell microscopy using fluorescent calcium indicators has become an indispensable tool for characterizing the mechanisms underlying both homeostatic and pathogen-induced cellular calcium dynamics, but such imaging must be coupled with robust quantitative analysis. Further, calcium imaging is most powerful when paired with reductive studies targeting calcium-modulating proteins. The lack of specific inhibitors or agonists to directly target most pathogen-induced calcium signals precludes many of the approaches that have allowed for robust characterization of major eukaryotic cell calcium signaling mechanisms, such as ER Ca2+ release by inositol triphosphate receptors. Given this, we sought to develop quantitative imaging pipelines tailored for the characterization of pathogen-induced calcium signals. Using rotavirus as a prototypical calcium-modulating pathogen, we developed and optimized a suite of computational tools for automated quantitation of both intra- and inter-cellular calcium signals detected via live-cell imaging of infected epithelial monolayers expressing genetically encoded calcium indicators. Using recombinant strains of rotavirus that express fluorescent markers, we developed a system that allows for automated detection of rotavirus-infected cells and normalization of signals to infectivity. All tools were built in ImageJ, making them freely available and adaptable across operating systems and microscope setups. These tools required minimal active time from the user and allowed for the extraction of signal parameters previously unquantifiable, increasing the speed and breadth of characterization.

Electrocardiogram (ECG) interpretation is fundamental for cardiac diagnosis. Machine learning has proven strong performance in ECG analysis but models often face challenges when deployed in real-world settings due to difficulties adapting to new environments. Recent progress, particularly through training on large, diverse datasets across various ECG-related tasks, has led to better model generalization. However, many cardiac conditions, particularly rare ones, remain scarce in existing datasets, leading to diminished model performance in these critical cases. This underrepresentation remains a major hurdle, as acquiring high-quality annotations for rare conditions is time-consuming and demands substantial clinical expertise and resources. In this paper, we investigate whether this issue can be addressed with the help of Large Vision-Language Models (VLMs) and In-Context Learning (ICL). We propose leveraging the inherent capabilities of VLMs, aided with only a few task-specific examples in the prompt to interpret ECG images in data-scarce environments, thus reducing reliance on extensive labeled data for fine-tuning. We introduce a real-world use case involving Brugada syndrome and evaluate the performance of pre-trained VLMs, comparing them to state-of-the-art ECG machine learning models. Results show that VLMs achieve competitive accuracy and, in data-constrained scenarios, outperform existing methods without requiring any updates to model weights. We further analyze the role of prompt engineering and input representation in influencing model performance. Our findings suggest that VLMs could serve as an alternative to address rare cardiac conditions, which are frequently overlooked because of data scarcity, positioning them as valuable assets for screening and triage-oriented analysis.

Superior hypophyseal artery (SHA) aneurysms are occasionally associated with infraoptic anterior cerebral artery (ACA) variants, characterized by the A1 segment coursing beneath the optic nerve.1,2 A 69-year-old female presented with an 8 mm right SHA aneurysm, and subsequent angiography demonstrated a Wong Type I infraoptic ACA, resulting in a dual arterial supply to the aneurysm.3 Due to this anatomy, dual overlapping Pipeline Shield flow diverters were deployed. However, one-year follow-up demonstrated persistent aneurysm flowing from continued infraoptic ACA inflow. Microsurgical clipping via supraorbital eyebrow craniotomy was performed, placing two mini-clips at the aneurysm-infraoptic ACA junction to eliminate anterograde filling while preserving retrograde perfusion. Complete exclusion with normal A1 patency was confirmed with intraoperative ICG angiography, fluorescein angiography, and formal cerebral angiography. Six-month imaging demonstrated persistent occlusion. Infraoptic ACA variants resulting in dual blood supply may necessitate combined embolization and flow diversion or microsurgical approaches when flow diversion alone proves inadequate. Institutional review board (IRB) approval was obtained and written informed consent was obtained from the patient for publication of this case video and accompanying images.

Settlement and early post-settlement survival of fishes are important processes on coral reefs and represent a major demographic bottleneck. While numerous studies have quantified habitat selection of recently-settled fish in terms of benthic composition or physical complexity, few have considered the potential role of habitat colouration. Using image-based analyses to identify key colours and benthic features that create these colours, we explored how recently-settled reef fish select various habitats with respect to their own colouration and habitat colour composition. We conducted visual surveys and took standardized photographs of the settlement locations of four species of damselfishes (f: Pomacentridae) at Lizard Island, Australia; two of which are bright yellow in appearance (Pomacentrus amboinensis and P. moluccensis) and two that are dark coloured (P. adelus and P. chrysurus). Each species occupied settlement habitats that had distinct colour profiles when compared to the wider reef environment and the other species examined. When species were grouped according to body colouration (dark vs yellow), there were also significant differences in settlement habitat colouration. These patterns appear to be driven primarily by the presence or absence of live coral, mainly Acropora spp., within their settlement location which add a range of unique hues not typical of other substrata. Thus, we provide evidence that habitat colouration, in addition to other physical and environmental attributes, may be important for shaping settlement patterns in coral reef fishes. If so, rapidly changing benthic compositions on coral reefs driven by the loss of live coral may ultimately favour certain coloured species that reside more successfully on drabber habitats which are becoming more common on contemporary reefs.

Tanorexia, commonly described as tanning dependence or tanning addiction, refers to a compulsive pattern of ultraviolet (UV)-seeking behavior in which individuals continue to tan despite awareness of, or exposure to, significant dermatologic and systemic risks. Rather than representing a simple aesthetic preference, tanorexia is increasingly understood as a psychodermatologic and biopsychosocial phenomenon involving reward-related neurobiological pathways, body-image concerns, genetic susceptibility, and psychiatric comorbidities. Indoor tanning is particularly concerning,because UV-emitting tanning devices are established carcinogenic exposures, and early initiation of sunbed use has been associated with a substantially increased risk of melanoma. Emerging evidence also suggests that excessive tanning shares clinical and behavioral features with substance-use and behavioral addictive disorders, including craving, impaired control, reinforcement, and continued use despite harm. For dermatologists, compulsive tanning should therefore be approached not only as a risk factor for photoaging and skin cancer, but also as a potential marker of underlying psychological vulnerability. Screening and assessment tools, including modified DSM criteria (mDSM), the modified CAGE questionnaire (mCAGE), the Structured Interview for Tanning Abuse and Dependence (SITAD), and the Behavioral Addiction Indoor Tanning Screener (BAITS), may help identify at-risk individuals and guide appropriate counseling or referral. This review evaluates the etiology, clinical manifestations, screening approaches, dermatologic consequences, and management implications of tanorexia, with emphasis on its relevance to dermatologic practice and public health prevention.

To investigate the leaching behavior of HMs from recycled industrial solid waste-based materials under cyclic-hydrostatic pressure and wet-dry cycling (C/C) caused by fluctuating groundwater tables in underground applications, a novel C/C environment simulation apparatus was designed. The temporal variations in HM leaching concentrations of red mud-flue gas desulfurization gypsum-based backfilling grout were compared under TCLP, C/C leaching, and constant-hydrostatic pressure leaching at 40, 80, 120, and 160 kPa. The effects of different leaching environments on the microstructure, mineral phase, and chemical characteristics of the grout were examined using SEM, MIP, FTIR, and XRD. Compared with the TCLP, C/C leaching under 160 kPa elevated the leachable concentrations of heavy metals. Specifically, the concentration of Pb rose from 1.6 ppb to 3.5 ppb, Cu from 18.9 ppb to 58.9 ppb, Cr from 25.6 ppb to 59.6 ppb, Cd from 0.43 ppb to 1.44 ppb, Mn from 0.25 ppm to 3.6 ppm, and As from 0.9 ppb to 48.2 ppb. The leaching concentrations of HMs showed strong correlations with those of structural elements (Fe, Na, S, and Si), especially with the Fe-S matrix. Combined with the chemical fractionation results, it indicates that C/C environment remobilizes part of the oxidizable fraction and a small amount of the reducible fraction of HMs. FTIR detected the penetration of leaching agent into the harden grout at a depth of 2 mm under C/C leaching at 160 kPa. MIP results revealed that C/C leaching significantly increased the total porosity and the proportion of macropores, demonstrating severe degradation of the hardened grout microstructure and enhanced leaching agent penetration. XRD results indicated obvious damage to the C(N)-(A)-S-H and AFt phases, while SEM images confirmed a substantial loss of surface compactness and integrity.

Huntington's disease (HD) is a genetic neurodegenerative disease characterized by striatum damage, which results in a number of uncontrollable muscle movements alongside intellectual and cognitive impairment. The progression of HD is accompanied by neuroinflammation, oxidative stress, and neuronal apoptosis. Resveratrol (RESV) is a naturally occurring compound known for its potent antioxidant and anti-inflammatory effects. RESV showed promising neuroprotective effects against Alzheimer's and Parkinson's disease. The current research aims to study the neuroprotective effects of RESV against 3-nitropropionic acid (3-NP)-induced HD. Forty adult male rats were divided equally into four groups as follows: Group 1- normal control group. Group 2- RESV (25 mg/kg/day, p.o) - treated rats. Group 3- rats treated with 3-NP (10 mg/kg/day, i.p). Group 4- rats treated with 3-NP (10 mg/kg/day, i.p) +RESV (25 mg/kg/day, p.o). The results showed that RESV alleviated the behavioral deficits observed in 3-NP treated rats. In addition, the histopathological images showed obvious improvement in RESV-treated rats. RESV activated the AMP-activated protein kinase (AMPK)-related autophagy pathway that resulted in neuroprotection and cell survival. Moreover, RESV showed anti-inflammatory and antioxidant effects by decreasing levels of inflammatory biomarkers including tumor necrosis factor (TNF)-α, nuclear factor kappa (NF-κ)-B, and interleukin (IL)-1β, alongside increasing neuronal antioxidant capacity by stimulating reduced glutathione (GSH), superoxide dismutase (SOD), and preventing lipid peroxidation. In conclusion, our study showed that RESV has a potent neuroprotective effect as evidenced by its ability to significantly alleviate biochemical and behavioral hallmarks of HD.

The diagnosis of normal-pressure hydrocephalus (NPH) is often complicated due to deficiencies of the objective measures currently used after test drainage of CSF. We used Arterial Spin Labeled Magnetic Resonance Imaging (ASL-MRI)-a novel, simplified, completely non-invasive, radiation-free method-to measure global cerebral blood flow (CBF) before and after performing a large-volume lumbar puncture (LVLP) in patients suspected of NPH. We compared baseline ASL-CBF in 20 patients (65-91 years old, mean: 75 years; 11 men) with history of recurrent falls from unsteady gait, urinary incontinence, cognitive decline, and ventriculomegaly (Evans index >0.30). After LVLP under fluoroscopy draining 20-53 mL of CSF we measured ASL-CBF and compared the cerebral perfusion with baseline values for whole brain, predefined cortical regions, deep grey nuclei, and periventricular white matter. Correlation was assessed with changes in gait speed and balance, neuropsychology testing and urinary incontinence. Post-LVLP all patients had significant increase in global ASL-CBF with mean values rising from 39 to 45 mL/100g/min (p <0.01). CBF enhancement was notable in gray matter regions, thalamus and periventricular frontal white matter. Draining &#x2264;40 mL of CSF resulted on average CBF increase of 0.9 mL/100g/min compared with 5.2 mL/100g/min after draining 50 mL of CSF (p <0.01) indicating a dose-response relationship whereby draining <40 mL of CSF may not be adequate to diagnose NPH. We confirmed the occurrence of CBF hypoperfusion in NPH. Linear mixed-effects model for regional blood flow analysis confirmed consistent enhancement of cerebral perfusion in all evaluated regions post-lumbar puncture. Exploratory analysis to correlate baseline CBF with the magnitude of change post-lumbar puncture revealed a negative correlation (Pearson r = -0.819 p = 0.000) indicating that patients with lower baseline CBF exhibited larger increases in perfusion after CSF drainage. There was a positive correlation between enhancement of CBF and improvement of gait speed and balance. Using ASL-MRI we have demonstrated that global cerebral hypoperfusion is a constant feature of NPH that improves with CSF drainage. As a result, the clinical diagnosis of NPH can be greatly simplified using ASL-MRI.

Classic infrared (IR) microscopy is limited by the diffraction limit, which obscures subcellular heterogeneity, and by the complex overlap of vibrational bands, which complicates precise molecular assignment. This study presents a comprehensive "IR map of the cell" that provides a standardized framework for label-free chemical identification of subcellular compartments across various human cell lines. By integrating Fourier transform infrared (FTIR) spectroscopy with submicron-resolution optical photothermal infrared (OPTIR) microscopy (&#x223c;0.3 &#x3bc;m), the cellular landscape was mapped with improved spatial specificity beyond that achievable by conventional FTIR imaging. Advanced chemometric tools were employed to segment the nucleus, cytoplasm, and regions rich in lipids and glycogen, each characterized by a definitive "IR barcode". Furthermore, in silico modeling validated spectral assignments by simulating cellular fingerprints from reference biocompounds, while detailed spectroscopic characterization of subcellular compartments defined marker bands for proteins, lipids, carbohydrates, and nucleic acids. The modeling supported the interpretation that experimental spectra can be approximated as a linear combination of biomolecular classes, helping to constrain spectral band overlap and refine the proposed "IR barcode" for cellular identification. The developed IR map provides a robust, standardized foundation for the label-free interpretation of cellular chemistry. This study demonstrates the utility of IR microscopy as a powerful diagnostic and analytical tool for monitoring metabolic shifts and cellular status at the micrometric level.

Separation surgery has emerged as a key surgical strategy for metastatic spinal cord compression (MSCC), aiming to create a circumferential decompressive margin that allows safe delivery of postoperative radiotherapy. However, despite its widespread adoption, the clinical value of objectively confirming separation on early postoperative magnetic resonance imaging (MRI) remains unclear. To evaluate whether separation success confirmed by MRI at 2-3 weeks postoperatively is associated with improved neurological recovery, functional outcomes, and survival in patients with MSCC. Retrospective cohort study. Fifty-nine patients who underwent posterior separation surgery for MSCC between 2020 and 2023 were included. All patients underwent metal artifact-reduced MRI at three weeks postoperatively and were classified into a separation group (Group S; n&#x202f;=&#x202f;26, 44.1%) or a non-separation group (Group NS; n&#x202f;=&#x202f;33, 55.9%) based on MRI findings. Primary outcomes included neurological recovery (motor grade and ambulation status), overall survival, and length of hospital stay. Secondary outcomes included radiologic parameters (Bilsky grade and Spinal Instability Neoplastic Score), postoperative complications, and radiotherapy administration. Successful separation was defined as a cerebrospinal fluid margin &#x2265;2 mm between the tumor and spinal cord. Baseline characteristics were compared using appropriate parametric and nonparametric tests. Logistic regression was used to identify predictors of separation success. Survival outcomes were analyzed using the Kaplan-Meier method with log-rank testing and Cox proportional hazards regression. Baseline characteristics were comparable between groups except for Bilsky grade distribution. Group NS had a significantly higher proportion of Bilsky grade 3 lesions (28/33 vs. 16/26, p&#x202f;=&#x202f;0.041). Compared with Group NS, Group S demonstrated significantly higher postoperative motor grades (4.54 vs. 3.67, p&#x202f;=&#x202f;0.015), higher ambulation rates (92.3% vs. 69.7%, p&#x202f;=&#x202f;0.032), and shorter hospital stays (15.81 vs. 23.94 days, p&#x202f;=&#x202f;0.042). Overall survival was significantly longer in Group S (13.31 vs. 6.02 months, p&#x202f;=&#x202f;0.001). Logistic regression identified preoperative Bilsky grade as the only independent predictor of separation failure (OR&#x202f;=&#x202f;0.248, 95% CI: 0.069-0.888, p&#x202f;=&#x202f;0.032). Cox regression demonstrated that separation failure was associated with a 2.63-fold increased risk of mortality (p&#x202f;=&#x202f;0.006). Separation success confirmed on early postoperative MRI obtained 2-3 weeks after surgery was associated with early neurological recovery, higher postoperative ambulation rates, and prolonged survival in patients with MSCC. Despite technical limitations, early postoperative MRI provides a practical and objective means of assessing decompression adequacy and may support postoperative evaluation and treatment planning following separation surgery for MSCC.

In temporal lobe epilepsy (TLE), recurrent seizures can cause structural and functional disruptions within the temporal lobe and nearby language regions, resulting in neural reorganization. In this study, we leverage task-based functional magnetic resonance imaging (tb-fMRI) to understand how this language reorganization process varies based on the seizure location, the timing of seizure onset, and the chronicity of epileptic activity. 84 drug-resistant TLE patients who completed a tb-fMRI sentence completion task were included in this study. Analysis included group comparison of activation and comparison of the extent of reorganization in the whole brain and standard language regions of interest (ROI) level. We also measured the impact of the age of onset and the duration of epilepsy in the reorganization process. A significantly higher degree of language network reorganization is observed in left (L) TLE than in right (R) TLE at the whole brain level. At the regional level, a higher degree of reorganization is observed in temporal and frontal lobe ROIs in LTLE compared to RTLE, especially in the middle frontal gyrus and posterior temporal gyrus. The effects of age of onset and epilepsy duration were prominent in LTLE subjects at both the whole-brain and ROI levels. Patients with LTLE show greater changes in the middle frontal and posterior temporal regions, particularly in those with an earlier age of onset and longer disease duration. Identifying these patterns may assist in surgical planning and personalizing treatment to protect cognitive function.