The spatial and functional organization of the primate visual cortex is a fundamental problem in neuroscience. While recent computational frameworks like the Topographic Deep Artificial Neural Network (TDANN) have successfully modeled spatial organization in the ventral stream, the computational origins of the dorsal stream's distinct topographies, such as direction-selective maps in the middle temporal (MT) area, remain largely unresolved. In this work, we present a spatiotemporal TDANN to investigate whether MT topography is governed by the same universal principles. By training a 3D ResNet on naturalistic videos via a Momentum Contrast (MoCo) self-supervised paradigm alongside a biologically inspired spatial loss, we demonstrate the spontaneous emergence of brain-like direction maps and topological pinwheel structures. Crucially, we reveal that MT tuning properties, characterized by strong direction selectivity paired with a residual axial component, arise from a strict optimization trade-off between task-driven discriminative pressure and spatial regularization. The model's representations quantitatively match in vivo macaque MT physiological baselines, including direction selectivity index, circular variance, and pinwheel density. These findings unify the computational origins of the ventral and dorsal streams, establishing a general mechanism for cortical self-organization.

Quantitative MR relaxometry provides sensitive and reproducible measures of brain microstructure and is increasingly used in studies of ageing and neurodegenerative disease. However, normative brain atlases covering the adult lifespan remain scarce, especially from large, harmonized, multicenter datasets. This study aimed to establish lifespan-based normative T1 and T2 atlases to support individualized assessment and group-level comparisons. We retrospectively analyzed 947 healthy Han Chinese adults (421 males; median age 39 years; range 19-72) from 11 imaging centers in China. T1 and T2 maps were acquired with harmonized MP2RAGE and GRAPPATINI protocols, processed using a unified pipeline, and modeled with voxelwise mixed-effects regression including linear and quadratic age terms, sex, and random intercepts for site. Intersite reproducibility was further evaluated in three traveling subjects scanned at all centers. The atlases demonstrated region-specific age and sex effects in both white and gray matter. Intersite variability was minimal (intraclass correlation coefficients > 0.99 for T1 and > 0.90 for T2). Both T1 and T2 followed quadratic trajectories, decreasing in early adulthood, reaching minima in midlife, and increasing in later decades, with the strongest effects in cortical gray matter. Sex differences were most pronounced in parietal and callosal regions. These lifespan-based normative relaxometry atlases provide high-fidelity references for age- and sex-related microstructural changes in the healthy brain. They provide quantitative benchmarks for individualized profiling and group comparisons in Chinese adult populations, while validation in other ethnic populations remains necessary before broader application.

The ability to robustly maintain and update continuous variables is a hallmark of working memory. While classical continuous attractor networks suffer from severe fine-tuning fragility, standard artificial recurrent neural networks (RNNs) like GRUs and LSTMs typically fail to stably learn continuous manifolds, instead shattering the state space into discretized point attractors. To bridge this gap, we draw inspiration from divisive normalization, a canonical neural computation widely observed across cortical circuits, and propose the Recurrent Divisive Normalization Network (RDNN), a minimal and algebraically isolated model of dynamic division. Through dynamical systems analysis on canonical working memory tasks, we demonstrate that this biophysical constraint allows the network to converge to robust, high-fidelity slow manifolds. Furthermore, we analyze the gradient dynamics of divisive normalization during Backpropagation Through Time (BPTT), showing that it introduces an activity-dependent local gradient scaling. This scaling dampens parameter updates in highly active regimes, which empirically aligns with a significant self-compression of the network's effective rank, confining the recurrent dynamics to a tight, low-dimensional subspace while avoiding the optimization pathologies associated with explicit low-rank factorization. Finally, ablations demonstrate that while subtractive inhibition can maintain static memories, divisive normalization is mathematically essential to prevent manifold shattering under time-varying inputs. Our findings identify divisive normalization not merely as a biological artifact, but as a critical computational mechanism for learning high-fidelity continuous representations.

High-resolution functional MRI (fMRI) at 7 T enables sub-millimeter mapping of brain activity, but 3D multi-shot echo-planar imaging (EPI) is often affected by Nyquist ghost and fuzzy ripple artifacts that degrade image quality and may bias functional interpretation. Dual-polarity GRAPPA (DPG) is a robust strategy for correcting polarity-related artifacts, yet its practical implementation and systematic evaluation in 3D multi-shot EPI for high-resolution fMRI remain limited. Here, a practical dual-polarity reconstruction pipeline was implemented for 3D multi-shot EPI and evaluated in phantom and in vivo experiments at 7 T. The method uses a two-step strategy in which acquired RO+/RO- lines are first replaced with phase-corrected data and the remaining missing samples are subsequently reconstructed with conventional GRAPPA. In the 3D EPI setting, kz-extended kernels were used to account for slice-/partition-dependent phase behavior. Relative to linear phase correction (LPC), the proposed implementation consistently reduced Nyquist ghost and fuzzy ripple artifacts across imaging protocols. In sub-millimeter resting-state fMRI, it improved temporal SNR in artifact-prone temporal-lobe regions. In sub-millimeter auditory task fMRI, it significantly reduced false-positive activations relative to LPC while showing good repeatability across repeated runs. These results provide a systematic evaluation of dual-polarity reconstruction in high-resolution 3D multi-shot fMRI and demonstrate its practical value for improving image fidelity and activation specificity in this setting.

In recent years, deep-learning algorithms have been widely adopted for medical-image registration owing to their high speed and accuracy, yet most works focus on scalar images. Diffusion MRI (dMRI) is primarily used to depict white-matter architecture in the brain. Because each voxel contains measurements acquired under multiple diffusion directions and b-values, dMRI constitutes a high-dimensional imaging modality. Existing techniques often neglect the directional information, particularly in regions with complex fiber crossings, compromising the alignment of white matter structures between individuals and potentially affecting subsequent analyses like tractography. Aligning Fiber Orientation Distribution Functions (fODF), which represent the distribution of fiber orientations per voxel, is a promising strategy to leverage directional information. This approach facilitates more anatomically precise correspondence of white matter pathways compared to methods relying solely on scalar information. This study proposes ODDRnet, an end-to-end, unsupervised deep-learning framework for nonlinear registration of dMRI data. ODDRnet directly aligns high-dimensional fODF derived from raw dMRI signals. It predicts a dense deformation field which is used to spatially warp the fODFs while appropriately reorienting the directional information, ensuring accurate alignment. ODDRnet was evaluated on multiple public and private datasets and compared with several state-of-the-art methods. Experimental results demonstrated that ODDRnet achieved superior macroscopic accuracy, with a mean increase of 0.02 in the tract dice and a reduction of 0.31 mm in tract distance. Furthermore, ODDRnet showed robust generalization across diverse ethnicities, age groups, health conditions, and imaging protocols, confirming its adaptability for various dMRI registration tasks.

Cognitive deficits in schizophrenia are a major contributor to poor functional outcomes and everyday functioning, making them a promising therapeutic target. Recent years have witnessed a dramatic increase in the use of digital interventions, such as game-based therapy, targeting various domains of cognition to treat mental disorders. Game-based digital interventions have been suggested to have therapeutic value in health care for people with schizophrenia. To support this idea, a novel, online training program (Komori Life) that targets cognitive deficits in schizophrenia was tested for feasibility of use and initial efficiency. Inpatients with schizophrenia were randomized to complete 20 sessions of either Komori Life (N = 40 completers) or treatment as usual (N = 40 completers). Cognitive and clinical assessments were performed at enrollment and after completion of the training intervention for all patients. In addition, 32 healthy volunteers were recruited as controls, and an eye-tracking paradigm was employed to assess attentional biases to emotional information before and after game intervention for all subjects. The results showed that there were no group differences in cognitive or clinical assessments at baseline between the two patient groups. After game training, there were still no group × time interactions on cognitive or clinical assessment scores. Regarding eye movement measurements, both patient groups showed increased attention to threatening stimuli compared to healthy controls in terms of attentional maintenance at baseline. After game training, the game training group revealed greater improvement in attentional bias towards threatening scenes (decreased percentage of total duration and percentage of total fixations towards threatening stimuli) relative to the treatment as usual group. Moreover, our results partially indicated that training effectiveness was associated with cognitive improvement and that heightened attentional maintenance to threats was associated with worse cognitive performance. This study provides initial evidence that a remote, online cognitive training program is feasible and effective in improving cognitive function in schizophrenia. This form of training may serve as a complementary therapy to existing psychiatric care. Clinical trial registration: the trial is registered at http://www.chictr.org.cn , identifier ChiCTR2100048403.

Purpose We aim to implement a 3D DWI sequence and show its usage on patients with new ischemic lesions. Materials and Methods The proposed 3D DWI sequence was implemented by integrating second-order gradient moment nulling (M2) and cardiac motion synchronization (Sync). All data were acquired on a 3T MAGNETOM Prisma scanner (Siemens Healthcare, Erlangen, Germany) using a 64 channel head and neck coil. 21 healthy volunteers underwent 3D DWI scans at 0.9 mm isotropic resolution using four motion compensation methods for comparison: no compensation (M0), M2 only, Sync only and the proposed M2+Sync method. 2D phase variation maps with different motion compensation methods were also acquired for one subject to illustrate the mechanism of the proposed method. A ghost-to-signal ratio (GSR) and blurring index was defined and compared among the four methods with repeated measures ANOVA and Tukey’s test. 3D DWI was compared with 2D DWI for ADC quantification. Image quality and ischemic lesion conspicuity were evaluated with 12 patients after endovascular treatment. >Results Whole brain 3D DWI was achieved at 0.9 mm isotropic resolution within 5 minutes using the proposed sequence. M2+Sync achieved the lowest level of GSR and blurring along the slice direction. ADC quantification showed no statistically significant difference between M2+Sync compared to 2D DWI. 3D DWI showed similar image quality, higher lesion conspicuity and counts compared to 2D DWI. Conclusion Direct 3D DWI can be achieved by the combination of second order gradient moment nulling and cardiac synchronization. ### Competing Interest Statement The authors have declared no competing interest.

Uncovering the fundamental neural correlates of biological intelligence, developing mathematical models, and conducting computational simulations are critical for advancing new paradigms in artificial intelligence (AI). In this study, we implemented a comprehensive visual decision-making model that spans from visual input to behavioral output, using a neural dynamics modeling approach. Drawing inspiration from the key components of the dorsal visual pathway in primates, our model not only aligns closely with human behavior but also reflects neural activities in primates, and achieving accuracy comparable to convolutional neural networks (CNNs). Moreover, magnetic resonance imaging (MRI) identified key neuroimaging features such as structural connections and functional connectivity that are associated with performance in perceptual decision-making tasks. A neuroimaging-informed fine-tuning approach was introduced and applied to the model, leading to performance improvements that paralleled the behavioral variations observed among subjects. Compared to classical deep learning models, our model more accurately replicates the behavioral performance of biological intelligence, relying on the structural characteristics of biological neural networks rather than extensive training data, and demonstrating enhanced resilience to perturbation.

Background: The use of shorter TR and finer atlases in rs-fMRI can provide greater detail on brain function and anatomy. However, there is limited understanding of the effect of this combination on brain network properties. Methods: A study was conducted with 20 healthy young volunteers who underwent rs-fMRI scans with both shorter (0.5s) and long (2s) TR. Two atlases with different degrees of granularity (90 vs 200 regions) were used to extract rs-fMRI signals. Several network metrics, including small-worldness, Cp, Lp, Eloc, and Eg, were calculated. Two-factor ANOVA and two-sample t-tests were conducted for both the single spectrum and five sub-frequency bands. Results: The network constructed using the combination of shorter TR and finer atlas showed significant enhancements in Cp, Eloc, and Eg, as well as reductions in Lp and gamma in both the single spectrum and subspectrum (p < 0.05, Bonferroni correction). Network properties in the 0.082-0.1 Hz frequency range were weaker than those in the 0.01-0.082 Hz range. Conclusion: Our findings suggest that the use of shorter TR and finer atlas can positively affect the topological characteristics of brain networks. These insights can inform the development of brain network construction methods.

ObjectivesIn this preregistered study, we investigated the beneficial effects of music-based casual video game training on the depression, anxiety and stress symptoms in a cohort of young individuals with subthreshold depression and the underlying mechanisms.MethodsThe study included 56 young individuals (18–26 years of age) with subthreshold or mild depression based on the Beck Depression Inventory-II (BDI-II) scores between 14 and 19. They were randomly assigned into the experimental group (n = 28) or the control group (n = 28). The experimental group underwent music-based casual video game training for 4 weeks. During the same time, the control group participants conducted daily life activities without any intervention. The study participants in the two groups were analyzed using the Depression Anxiety and Stress Scale (DASS-21) during the baseline before the intervention, as well as DASS-21, Positive and negative Affect Scale (PANAS), General Self-efficacy Scale (GSES), and the Emotional Regulation Questionnaire (ERQ) twice a week during the 4 weeks of intervention.ResultsThe depression, anxiety, and stress symptoms were significantly reduced in the experimental group participants after 4 weeks of music-based video game training compared with the control group. The DAS scores in the experimental group were alleviated in the third and fourth weeks of training compared with the control group. Moreover, analysis using the general linear model demonstrated that the number of training weeks and self-efficacy were associated with significant reduction in depression, anxiety and stress. Furthermore, our results demonstrated that self-efficacy was correlated with positive emotion and emotional regulation.ConclusionOur study showed that music-based casual video game training significantly decreased depression, anxiety, and stress in the young individuals with subthreshold depression by enhancing self-efficacy.

This study aimed to establish a non-invasive radiomics model based on computed tomography (CT), with favorable sensitivity and specificity to predict EGFR mutation status in GGO-featured lung adenocarcinoma subsequently guiding the administration of targeted therapy. Clinical-pathological information and preoperative CT images of 636 lung adenocarcinoma patients (464, 100, and 72 in the training, internal, and external validation sets, respectively) that underwent GGO lesions resection were included. A total of 1476 radiomics features were extracted with gradient boosting decision tree (GBDT). The established radiomics model containing 102 selected features showed an encouraging discrimination performance of EGFR mutation status (mutant or wild type), and the predictive ability was superior to that of the clinical model (AUC: 0.838 vs. 0.674, 0.822 vs. 0.730, and 0.803 vs. 0.746 for the training, internal validation, and external validation sets, respectively). The combined radiomics plus clinical model showed no additional benefit over the radiomics model in predicting EGFR status (AUC: 0.846 vs. 0.838, 0.816 vs. 0.822, and 0.811 vs. 0.803, respectively, in three cohorts). Uniquely, this model was validated in a cohort of lung adenocarcinoma patients who have undertaken adjuvant EGFR-TKI treatment and harbored unresected GGOs during the medication, leading to a significantly improved potency of EGFR-TKIs (response rate: 25.9% vs. 53.8%, p = 0.006; before and after prediction, respectively). This presented radiomics model can be served as a non-invasive and time-saving approach for predicting the EGFR mutation status in lung adenocarcinoma presenting as GGO. • We developed a GGO-specific radiomics model containing 102 radiomics features for EGFR mutation status differentiation. • An AUC of 0.822 and 0.803 in the internal and external validation cohorts, respectively, were achieved. • The radiomics model was utilized in clinical translation in an adjuvant EGFR-TKI treatment cohort with unresected GGOs. A significant improvement in the potency of EGFR-TKIs was achieved (response rate: 25.9% vs. 53.8%, p = 0.006; before and after prediction).

Objective To evaluate changes in morphology of the cesarean scar and uterus between one and two years after cesarean section using high-resolution, three dimensional T2-weighted sampling perfection with application optimized contrast using different flip angle evolutions Magnetic Resonance Imaging (3D T2w SPACE MRI). Methods This prospective study was performed to investigate morphological changes in the cesarean scars and uterus from one to two years after cesarean section using high-resolution, 3D T2w SPACE MRI. The healthy volunteers having no childbearing history were recruited as the controls. All data were measured by two experienced radiologists. All data with normal distribution between the one-year and two-year groups were compared using a paired-sample t test or independent t test. Results Finally, 46 women took a pelvic MR examination one year after cesarean section, and a subset of 15 completed the same examination again after two years of cesarean section. Both the uterine length and the anterior wall thickness after two years of cesarean section (5.75 ± 0.46 and 1.45 ± 0.35 cm) were significantly greater than those measured at one year (5.33 ± 0.59 and 1.25 ± 0.27 cm) (t = -2.363 and -2.175, P= 0.033 and 0.048). No significant difference was shown in myometrial thickness two years after cesarean section (1.45 ± 0.35 cm) with respect to the control group (1.58 ± 0.21 cm, P = 0.170). Nine women who underwent MRI twice were considered to have scar diverticula one year after cesarean section, and still had diverticula two years after cesarean section. The thickness, height, and width of the uterine scar showed no significant change from one to two years (all P > 0.05). Conclusions 3D T2w SPACE MRI provides overall morphologic details and shows dynamic changes in the scar and the uterus between one and two years after cesarean section. Scar morphology after cesarean section reached relatively stable one year after cesarean section, and uterine morphology was closer to normal two years after cesarean section.

Background:Magnetic resonance (MR) images generated by different scanners generally have inconsistent contrast properties, making it difficult to perform a combined quantitative analysis of images from a range of scanners. In this study, we aimed to develop an automatic brain image segmentation model to provide a more reliable analysis of MR images taken with different scanners.Methods:The spatially localized atlas network tiles-27 (SLANT-27) deep learning model was used to train the automatic segmentation module, based on a multi-center dataset of 1,917 three-dimensional (3D) T1-weighted MR images. Subsequently, a framework called Qbrain, consisting of a new generative adversarial network (GAN) image transfer module and the SLANT-27 segmentation module, was developed. Another 3D T1-weighted MRI interscan dataset of 48 participants who were scanned in 3 MRI scanners (1.5T Siemens Avanto, 3T Siemens Trio Tim, and 3T Philips Ingenia) on the same day was used to train and test the Qbrain model. Volumetric T1-weighted images were processed with Qbrain, SLANT-27, and FreeSurfer (FS). The automatic segmentation reliability across the scanners was assessed using test-retest variability (TRV).Results:The reproducibility of different segmentation methods across scanners showed a consistent trend in the greater reliability and robustness of QBrain compared to SLANT-27 which, in turn, showed greater reliability and robustness compared to FS. Furthermore, when the GAN image transfer module was added, the mean segmentation error of the TRV of the 3T Siemens vs. 1.5T Siemens, the 3T Philips vs. 1.5T Siemens, and the 3T Siemens vs. 3T Philips scanners was reduced by 1.57%, 2.01%, and 0.56%, respectively. In addition, the segmentation model improved intra-scanner variability (0.9-1.67%) compared with that of FS (2.47-4.32%).Conclusions:The newly developed QBrain method combined with GAN image transfer module and a SLANT-27 segmentation module was shown to improve the reliability of whole-brain automatic structural segmentation results across multiple scanners, thus representing a suitable alternative quantitative method of comparative brain tissue analysis for individual patients.

Till March 31st, 2021, the coronavirus disease 2019 (COVID-19) had reportedly infected more than 127 million people and caused over 2.5 million deaths worldwide. Timely diagnosis of COVID-19 is crucial for management of individual patients as well as containment of the highly contagious disease. Having realized the clinical value of non-contrast chest computed tomography (CT) for diagnosis of COVID-19, deep learning (DL) based automated methods have been proposed to aid the radiologists in reading the huge quantities of CT exams as a result of the pandemic. In this work, we address an overlooked problem for training deep convolutional neural networks for COVID-19 classification using real-world multi-source data, namely, the data source bias problem. The data source bias problem refers to the situation in which certain sources of data comprise only a single class of data, and training with such source-biased data may make the DL models learn to distinguish data sources instead of COVID-19. To overcome this problem, we propose MIx-aNd-Interpolate (MINI), a conceptually simple, easy-to-implement, efficient yet effective training strategy. The proposed MINI approach generates volumes of the absent class by combining the samples collected from different hospitals, which enlarges the sample space of the original source-biased dataset. Experimental results on a large collection of real patient data (1,221 COVID-19 and 1,520 negative CT images, and the latter consisting of 786 community acquired pneumonia and 734 non-pneumonia) from eight hospitals and health institutions show that: 1) MINI can improve COVID-19 classification performance upon the baseline (which does not deal with the source bias), and 2) MINI is superior to competing methods in terms of the extent of improvement.

The molecular subtyping of diffuse gliomas is important. The aim of this study was to establish predictive models based on preoperative multiparametric MRI. A total of 1016 diffuse glioma patients were retrospectively collected from Beijing Tiantan Hospital. Patients were randomly divided into the training (n = 780) and validation (n = 236) sets. According to the 2016 WHO classification, diffuse gliomas can be classified into four binary classification tasks (tasks I–IV). Predictive models based on radiomics and deep convolutional neural network (DCNN) were developed respectively, and their performances were compared with receiver operating characteristic (ROC) curves. Additionally, the radiomics and DCNN features were visualized and compared with the t-distributed stochastic neighbor embedding technique and Spearman’s correlation test. In the training set, areas under the curves (AUCs) of the DCNN models (ranging from 0.99 to 1.00) outperformed the radiomics models in all tasks, and the accuracies of the DCNN models (ranging from 0.90 to 0.94) outperformed the radiomics models in tasks I, II, and III. In the independent validation set, the accuracies of the DCNN models outperformed the radiomics models in all tasks (0.74–0.83), and the AUCs of the DCNN models (0.85–0.89) outperformed the radiomics models in tasks I, II, and III. DCNN features demonstrated more superior discriminative capability than the radiomics features in feature visualization analysis, and their general correlations were weak. Both the radiomics and DCNN models could preoperatively predict the molecular subtypes of diffuse gliomas, and the latter performed better in most circumstances. • The molecular subtypes of diffuse gliomas could be predicted with MRI. • Deep learning features tend to outperform radiomics features in large cohorts. • The correlation between the radiomics features and DCNN features was low.

我国基层阴道镜医生诊断表现不佳是宫颈癌筛查体系中的难点和痛点.开展阴道镜培训是提高基层医生诊断水平的有效方式,但由于我国基层医生人数众多且操作经验较少,专家线下授课的培训模式进展较为缓慢.随着计算机技术的发展和临床大数据的积累,研发一款科学、权威、实用、富有趣味性的智能化培训软件有望提升阴道镜医生的诊断水平.本研究旨在介绍研发智能阴道镜培训软件的研究意义、研究内容和相关研究进展,并探讨其应用的优势及现实挑战.

Cervical cancer causes the fourth most cancer-related deaths of women worldwide. Early detection of cervical intraepithelial neoplasia (CIN) can significantly increase the survival rate of patients. World Health Organization (WHO) divided the CIN into three grades (CIN1, CIN2 and CIN3). In clinical practice, different CIN grades require different treatments. Although existing studies proposed computer aided diagnosis (CAD) systems for cervical cancer diagnosis, most of them are fail to perform accurate separation between CIN1 and CIN2/3, due to the similar appearances under colposcopy. To boost the accuracy of CAD systems, we construct a colposcopic image dataset for GRAding cervical intraepithelial Neoplasia with fine-grained lesion Description (GRAND). The dataset consists of colposcopic images collected from 8,604 patients along with the pathological reports. Additionally, we invite the experienced colposcopist to annotate two main clues, which are usually adopted for clinical diagnosis of CIN grade, i.e., texture of acetowhite epithelium (TAE) and appearance of blood vessel (ABV). A multi-rater model using the annotated clues is benchmarked for our dataset. The proposed framework contains several sub-networks (raters) to exploit the fine-grained lesion features TAE and ABV, respectively, by contrastive learning and a backbone network to extract the global information from colposcopic images. A comprehensive experiment is conducted on our GRAND dataset. The experimental results demonstrate the benefit of using additional lesion descriptions (TAE and ABV), which increases the CIN grading accuracy by over 10%. Furthermore, we conduct a human-machine confrontation to evaluate the potential of the proposed benchmark framework for clinical applications. Particularly, three colposcopists on different professional levels (intern, in-service and professional) are invited to compete with our benchmark framework by investigating a same extra test set—our framework achieves a comparable CIN grading accuracy to that of a professional colposcopist.

There is close relationship between lenticulostriate arteries (LSAs) and lacunar infarctions (LIs) of the basal ganglia. The study aims to visualize the LSAs using high-resolution vessel wall imaging (VWI) on 3T system and explore the correlation between LSAs and LIs. Fifty-six patients with LIs in basal ganglia, and 44 age-matched control patients were enrolled and analyzed retrospectively. The raw VWI images were reformatted into coronal slices in minimum intensity projection for further observation of LSAs. The risk factors of LIs in basal ganglia were analyzed by univariate and multivariate logistic regression. The correlation and linear regression analysis between the LSAs and LIs, ipsilateral MCA-M1 plaques were investigated. The total number (p < 0.01) and length (p < 0.01) of LSAs were statistically different between basal ganglias with and without LIs. The total number of LSAs and ipsilateral MCA-M1 plaques were independently related to LIs in basal ganglias. The mean length of LSAs were negatively correlated with number (r = − 0.33, p = 0.002) and volume (r = − 0.37, p = 0.001) of LIs. Age, drinking history, and mean length of LSAs were associated with LI occurrence in basal ganglia, and mean length of LSAs was correlated with larger volume of LIs. Number of LSA reduction and ipsilateral MCA-M1 plaques were associated with the presence of LIs in basal ganglias. Age increasing, drinking history, and shorter LSAs were correlated with the increasing of LIs. • Patients with LIs tend to have shorter LSAs. • The characteristics of LSAs and ipsilateral MCA-M1 plaques are associated with LIs in basal ganglias. • Age, drinking history, and mean length of LSAs are correlated with LI features in basal ganglias.

Background Multisite studies can considerably increase the pool of normally aging individuals with neurodegenerative disorders and thereby expedite the associated research. Understanding the reproducibility of the parameters of related brain structures-including the hippocampus, amygdala, and entorhinal cortex-in multisite studies is crucial in determining the impact of healthy aging or neurodegenerative diseases. Purpose To estimate the reproducibility of the fascinating structures by automatic (FreeSurfer) and manual segmentation methods in a well-controlled multisite dataset. Material and Methods Three traveling individuals were scanned at 10 sites, which were equipped with the same equipment (3T Prisma Siemens). They used the same scan protocol (two inversion-contrast magnetization-prepared rapid gradient echo sequences) and operators. Validity coefficients (intraclass correlations coefficient [ICC]) and spatial overlap measures (Dice Similarity Coefficient [DSC]) were used to estimate the reproducibility of multisite data. Results ICC and DSC values varied substantially among structures and segmentation methods, and values of manual tracing were relatively higher than the automated method. ICC and DSC values of structural parameters were greater than 0.80 and 0.60 across sites, as determined by manual tracing. Low reproducibility was observed in the amygdala parameters by automatic segmentation method (ICC = 0.349-0.529, DSC = 0.380-0.873). However, ICC and DSC scores of the hippocampus were higher than 0.60 and 0.65 by two segmentation methods. Conclusion This study suggests that a well-controlled multisite study could provide a reliable MRI dataset. Manual tracing of volume assessments is recommended for low reproducibility structures that require high levels of precision in multisite studies.