Fiber orientation distribution (FOD) is an advanced diffusion MRI modeling technique that represents complex white matter fiber configurations, and a key step for subsequent brain tractography and connectome analysis. Its reliability and accuracy, however, heavily rely on the quality of the MRI acquisition and the subsequent estimation of the FODs at each voxel. Generating reliable FODs from widely available clinical protocols with single-shell and low-angular-resolution acquisitions remains challenging but could potentially be addressed with recent advances in deep learning-based enhancement techniques. Despite advancements, existing methods have predominantly been assessed on healthy subjects, which have proved to be a major hurdle for their clinical adoption. In this work, we validate a newly optimized enhancement framework, FastFOD-Net, across healthy controls and six neurological disorders. This accelerated end-to-end deep learning framework enhancing FODs with superior performance and delivering training/inference efficiency for clinical use (60× faster comparing to its predecessor). With the most comprehensive clinical evaluation to date, our work demonstrates the potential of FastFOD-Net in accelerating clinical neuroscience research, empowering diffusion MRI analysis for disease differentiation, improving interpretability in connectome applications, and reducing measurement errors to lower sample size requirements. Critically, this work will facilitate the more widespread adoption of, and build clinical trust in, deep learning based methods for diffusion MRI enhancement. Specifically, FastFOD-Net enables robust analysis of real-world, clinical diffusion MRI data, comparable to that achievable with high-quality research acquisitions.

BACKGROUND AND OBJECTIVES:In this study, we examine the long-term changes in chronic lesion tissue (CLT) among patients with relapsing-remitting MS (RRMS), focusing on its impact on clinical and radiologic disease progression indicators. METHODS:The study involved 72 patients with multiple sclerosis with at least a 5-year follow-up. Annual assessments used 3D fluid-attenuated inversion recovery (FLAIR), precontrast and postcontrast 3D T1, and diffusion-weighted MRI. Lesion segmentation was conducted using iQ-MS software, while brain structures were segmented using AssemblyNet. Volumetric changes in CLT were tracked using a novel custom-designed pipeline that estimates longitudinal volumetric changes in CLT using serial MRI data. RESULTS:Throughout the follow-up period, the volume of CLT in the entire cohort increased continuously and steadily, averaging 7.75% ± 8.2% or 315 ± 465 mm³ per year. Patients with expanding CLT experienced significantly faster brain atrophy, affecting both white and gray matter, particularly in the brain's central area. Expanded CLT was also associated with higher and worsening Expanded Disability Status Scale (EDSS) scores, in contrast to the stable CLT group, where EDSS remained unchanged. Sample size calculation for a clinical trial investigating the effect of treatment on slow expansion of chronic lesions demonstrated that a relatively small cohort of patients with RRMS, ranging from 24 to 69 patients per arm, would be required. DISCUSSION:This study demonstrates that, over a period of up to 5 years, patient-specific enlargement of CLT, when present, progresses at a constant rate and significantly influences brain atrophy and disease progression. In addition, the study underscores CLT as a promising biomarker for RRMS progression and suggests the feasibility of smaller, targeted clinical trials to evaluate treatments aimed at reducing chronic lesion expansion.

Inflammation is becoming increasingly recognised as a core feature of dementia with evidence indicating that its role may vary and adapt across different stages of the neurodegenerative process. This study aimed to investigate whether the associations of high-sensitivity C-reactive protein (hs-CRP) with neuropsychological performance (verbal memory, executive function, processing speed) and cerebral white matter hyperintensities (WMHs) differed between older adults with subjective cognitive decline (SCD; n = 179) and mild cognitive impairment (MCI; n = 286). Fasting serum hs-CRP concentrations were grouped into low (<1.0 mg/L), moderate (1.0-3.0 mg/L), and high (>3.0-10.0 mg/L). Structural MRI scans were used to estimate WMH lesion volumes in the whole brain, as well as periventricular, deep white matter, and frontal regions. After adjusting for relevant demographic and clinical factors, multiple regression analyses revealed that in participants with SCD, high hsCRP concentrations were significantly associated with poorer executive function ((3[95% CI] = - .20[- .65, -.04], p = .025) and processing speed ((3[95% CI] = - .19[- .53, .00], p = .048). Exploratory analyses suggested that this effect may be specific to APOE-84 non-carriers only. There were no significant associations between hsCRP and neuropsychological outcomes in those with MCI. Hs-CRP was not associated with WMH volumes. Our findings suggest that hs-CRP may be involved in early disruptions to cerebral frontal-subcortical pathways, particularly in APOE-84 non-carriers, though this association may be independent of white matter lesions. In the earliest stages of cognitive decline where subjective complaints are paramount, addressing inflammation may offer potential benefits for supporting cognitive health.

•Previous studies have revealed that mutations in regulator of telomere elongation helicase 1 (RTEL1) can lead to idiopathic pulmonary fibrosis (IPF).•Here, we identified three novel mutations (NM_001283009.1: p.G201R, p.D266H and p.S798T) of RTEL1 in IPF patients. Functional studies suggested that these three variants disrupted the protein stability and nucleocytoplasmic distribution of the RTEL1 protein, which further decreased the expression of dyskerin and finally reduced the length of telomere and led to IPF and related disorders.•Our study may broaden the mutation and phenotype spectrum of RTEL1 deficiency, offering insights into the mechanisms underlying IPF.

There have been promising outcomes from the use of unilateral High-intensity focused ultrasound (HiFUS) thalamotomy in tremor-dominant Parkinson's disease. However, the reliability of this treatment has been questioned due to the high rate of tremor relapse. Authors have hypothesized that treatment failure is due to insufficient HiFUS lesion size, though detailed volumetric lesion analyses are lacking. To report the blinded tremor outcomes of unilateral HiFUS thalamotomy in tremor-dominant Parkinson's disease and correlate these outcomes with lesion characteristics, including the dentatorubrothalamic tract ablation overlap, which may provide valuable insights into the mechanisms behind tremor relapse and ultimately refine the optimal HiFUS target for tremor in Parkinson's disease. Retrospective review of consecutively treated tremor-dominant Parkinson's disease patients followed under a uniform protocol. Blinded tremor analysis was completed on pre- and post-operative videos. Patients were classified into two groups: 'responder' (≥50% improvement in Hand Tremor Score) or 'suboptimal responder' (<50% improvement in Hand Tremor Score) at the last follow-up. 17 patients with tremor-dominant Parkinson's disease underwent a unilateral HiFUS thalamotomy at our centre. Pre- and post-operative videos were available in 15 patients for analysis. Baseline median Hand Tremor Score was 11.0 (9.5-14.5), improving to 6.0 (1-13.5) over a median 24-month (3-36) follow-up period (P = 0.098). Seven patients had ≥50% improvement in Hand Tremor Score, while eight patients had <50% improvement in Hand Tremor Score at the last follow-up. At the final follow-up, the median change in Hand Tremor Score from baseline was 91% in responders, compared to an 8% increase in the suboptimal responders (P < 0.002). Levodopa daily dose equivalent increased by 17% in responders (P = 0.043), and the difference between responders and suboptimal responders was significant at the final follow-up (P = 0.024). There was a trend for larger lesions in the suboptimal responders, 157.5mm3 (113.4-215) compared to 119.7mm3 (111.6-237.6) in responders. Further, the dentatorubrothalamic core lesion overlap was greater in the suboptimal responders, 41.7% (40.5-48.8%), compared to the responders, 27.1% (13.8-39.3%), (P = 0.010), and was associated with a higher Hand Tremor Score at the final follow-up. We found that unilateral HiFUS thalamotomy in tremor-dominant Parkinson's Disease resulted in sustained tremor reduction in approximately 50% of patients but was also in the context of higher levodopa replacement. These favourable outcomes did not correlate with DRTT ablation overlap or lesion size, providing indirect evidence that the most efficacious HiFUS thalamic tremor target differs between essential tremor and tremor-dominant Parkinson's Disease.

Diffusion-weighted imaging (DWI) is a type of Magnetic Resonance Imaging (MRI) technique sensitised to the diffusivity of water molecules, offering the capability to inspect tissue microstructures and is the only in-vivo method to reconstruct white matter fiber tracts non-invasively. The DWI signal can be analysed with the diffusion tensor imaging (DTI) model to estimate the directionality of water diffusion within voxels. Several scalar metrics, including axial diffusivity (AD), mean diffusivity (MD), radial diffusivity (RD), and fractional anisotropy (FA), can be further derived from DTI to quantitatively summarise the microstructural integrity of brain tissue. These scalar metrics have played an important role in understanding the organisation and health of brain tissue at a microscopic level in clinical studies. However, reliable DTI metrics rely on DWI acquisitions with high gradient directions, which often go beyond the commonly used clinical protocols. To enhance the utility of clinically acquired DWI and save scanning time for robust DTI analysis, this work proposes DirGeo-DTI, a deep learning-based method to estimate reliable DTI metrics even from a set of DWIs acquired with the minimum theoretical number (6) of gradient directions. DirGeo-DTI leverages directional encoding and geometric constraints to facilitate the training process. Two public DWI datasets were used for evaluation, demonstrating the effectiveness of the proposed method. Extensive experimental results show that the proposed method achieves the best performance compared to existing DTI enhancement methods and potentially reveals further clinical insights with routine clinical DWI scans. The code of the proposed DirGeo-DTI is available at https://mri-synthesis.github.io/ .

The human brain is a complex and highly dynamic system, and our current knowledge of its functional mechanism is still very limited. Fortunately, with functional magnetic resonance imaging (fMRI), we can observe blood oxygen level-dependent (BOLD) changes, reflecting neural activity, to infer brain states and dynamics. In this paper, we ask the question of whether the brain states represented by the regional brain fMRI can be predicted. Due to the success of self-attention and the transformer architecture in sequential auto-regression problems (e.g., language modelling or music generation), we explore the possibility of the use of transformers to predict human brain resting states based on the large-scale high-quality fMRI data from the human connectome project (HCP). Current results have shown that our model can accurately predict the brain states up to 5.04 s with the previous 21.6 s. Furthermore, even though the prediction error accumulates for the prediction of a longer time period, the generated fMRI brain states reflect the architecture of functional connectome. These promising initial results demonstrate the possibility of developing generative models for fMRI data using self-attention that learns the functional organization of the human brain. Our code is available at: https://github.com/syf0122/brain_state_pred .

Understanding how spontaneous brain activity relates to stimulus-driven neural responses is a fundamental challenge in cognitive neuroscience. While task-based functional magnetic resonance imaging (fMRI) captures localized stimulus-evoked brain activation, its acquisition is costly, time-consuming, and difficult to scale across populations. In contrast, resting-state fMRI (rs-fMRI) is task-free and abundant, but lacks direct interpretability. We introduce Rest2Visual, a conditional generative model that predicts visually evoked fMRI (ve-fMRI) from resting-state input and 2D visual stimuli. It follows a volumetric encoder–decoder design, where multiscale 3D features from rs-fMRI are modulated by image embeddings via adaptive normalization, enabling spatially accurate, stimulus-specific activation synthesis. To enable model training, we construct a large-scale triplet dataset from the Natural Scenes Dataset (NSD), aligning each rs-fMRI volume with stimulus images and their corresponding ve-fMRI activation maps. Quantitative evaluation shows that the predicted activations closely match ground truth across standard similarity and representational metrics, and support successful image reconstruction in downstream decoding. Notably, the predicted maps preserve subject-specific structure, demonstrating the model's capacity to generate individualized functional surrogates. Our results provide compelling evidence that individualized spontaneous neural activity can be transformed into stimulus-aligned representations, opening new avenues for scalable, task-free functional brain modeling.

Diffusion MRI is sensitive to white matter changes in amyotrophic lateral sclerosis (ALS). The current study aimed to establish disease profiles across core motor pathways, and their relevance to clinical progression in ALS. Sixty-five participants (ALS = 47; Control = 18) were recruited for the study. White matter integrity of motor, somatosensory, and premotor subdivisions within the corticospinal tract and corpus callosum were quantified by fibre density, fibre-bundle cross-section, structural connectivity, and fractional anisotropy. Analyses focused on identifying diffusion metrics and tract profiles sensitive to ALS pathology, and their association with clinical progression. Reduced fibre density of the motor subdivision of the corpus callosum (CC) and corticospinal tract (CST) demonstrated best performance in classifying ALS from controls (area-under-curve: CCmotor = 0.81, CSTmotor = 0.76). Significant reductions in fibre density (CCmotor: p < 0.001; CSTmotor: p = 0.016), and structural connectivity (CCmotor: p = 0.008; CSTsomatosensory: p = 0.012) indicated presence of ALS pathology. Reduced fibre density cross-section significantly correlated with severity of functional impairment (ALSFRS-R; CCmotor: r = 0.52, p = 0.019; CSTmotor: r = 0.59, p = 0.016). The largest effect sizes were generally found for motor and somatosensory subdivisions across both major white matter bundles. Current findings suggest that ALS does not uniformly impact the corticospinal tract and corpus callosum. There is a preferential disease profile of neurodegeneration mainly impacting primary motor fibres. Microstructural white matter abnormality indicated presence of ALS pathology while macrostructural white matter abnormality was associated with severity of functional impairment. Quantification of white matter abnormality in corticospinal tract and callosal subdivisions holds translational potential as an imaging biomarker for neurodegeneration in ALS.

Background and ObjectivesIn this study, we examine the long-term changes in chronic lesion tissue (CLT) among patients with relapsing-remitting MS (RRMS), focusing on its impact on clinical and radiologic disease progression indicators.MethodsThe study involved 72 patients with multiple sclerosis with at least a 5-year follow-up. Annual assessments used 3D fluid-attenuated inversion recovery (FLAIR), precontrast and postcontrast 3D T1, and diffusion-weighted MRI. Lesion segmentation was conducted using iQ-MS software, while brain structures were segmented using AssemblyNet. Volumetric changes in CLT were tracked using a novel custom-designed pipeline that estimates longitudinal volumetric changes in CLT using serial MRI data.ResultsThroughout the follow-up period, the volume of CLT in the entire cohort increased continuously and steadily, averaging 7.75% +/- 8.2% or 315 +/- 465 mm(3) per year. Patients with expanding CLT experienced significantly faster brain atrophy, affecting both white and gray matter, particularly in the brain's central area. Expanded CLT was also associated with higher and worsening Expanded Disability Status Scale (EDSS) scores, in contrast to the stable CLT group, where EDSS remained unchanged. Sample size calculation for a clinical trial investigating the effect of treatment on slow expansion of chronic lesions demonstrated that a relatively small cohort of patients with RRMS, ranging from 24 to 69 patients per arm, would be required.DiscussionThis study demonstrates that, over a period of up to 5 years, patient-specific enlargement of CLT, when present, progresses at a constant rate and significantly influences brain atrophy and disease progression. In addition, the study underscores CLT as a promising biomarker for RRMS progression and suggests the feasibility of smaller, targeted clinical trials to evaluate treatments aimed at reducing chronic lesion expansion.

Diffusion-weighted imaging (DWI) is a magnetic resonance imaging (MRI) technique that enables non-invasive mapping of structural brain connectivity. Fiber orientation distribution (FOD) is a crucial intermediate model used to represent complex white matter fiber configurations within a voxel and serves as a key step for subsequent brain tractography and connectome analysis. However, obtaining reliable FOD estimates is particularly challenging due to the constraints of acquisition protocols, which often feature low spatial and angular resolutions in non-research-focused clinical settings. Current deep learning approaches could address these limitations, but they typically enhance FOD quality only for either low angular or low spatial resolution inputs. Here, we present a unified and robust deep learning framework, namely UFREE, for FOD enhancement from varied resolutions. UFREE can produce high quality FOD images that align closely with ground-truth data from multimodal analysis of FOD derivatives. This framework holds the potential to bring research-level FOD quality to routine neuroimaging acquired in the clinical environment.

BACKGROUND:The choroid plexus (CP) is increasingly recognised as a contributor to chronic inflammation in multiple sclerosis (MS). While CP enlargement is reported in early MS, its role in secondary progressive MS (SPMS) is poorly understood. We aimed to quantify CP volume in SPMS and compare it to relapsing-remitting MS (RRMS) and clinically isolated syndrome (CIS), and to assess associations with disease severity and progression. METHODS:CP volumes were manually segmented and normalised to intracranial volume in 121 MS patients (32 CIS, 49 RRMS, 39 SPMS). Age correction was applied using a healthy control cohort (n = 109). Cross-sectional and longitudinal analyses were performed to assess associations between baseline CP volume, ventricular size, lesion burden, and brain atrophy. RESULTS:CP volume increased significantly across MS phenotypes: SPMS patients showed 32 % higher CP volume than CIS (p < 0.001) and 25 % higher than RRMS (p < 0.001). CP enlargement in SPMS was independent of ventricular size, indicating distinct underlying mechanisms. While lesion burden was the primary determinant of brain atrophy in SPMS, longitudinal data revealed significant associations of baseline CP volume with chronic lesion expansion (r²=0.33) and brain volume loss (r²=0.51). CONCLUSION:CP enlargement is a progressive feature of MS, not driven by ventricular expansion. In SPMS, it may reflect ongoing inflammation contributing to tissue damage, supporting its role as a biomarker.

Abstract Background The global dementia prevalence is surging, necessitating research into contributing factors. We aimed to investigate the association between metabolic syndrome (MetS), its components, serum uric acid (SUA) levels, and dementia risk. Methods Our prospective study comprised 466,788 participants without pre-existing MetS from the UK Biobank. We confirmed dementia diagnoses based on the ICD-10 criteria (F00-03). To evaluate the dementia risk concerning MetS, its components, and SUA levels, we applied Cox proportional hazards models, while adjusting for demographic factors. Results Over a median follow-up of 12.7 years, we identified 6845 dementia cases. Individuals with MetS had a 25% higher risk of all-cause dementia (hazard ratio [HR] = 1.25, 95% confidence interval [CI] = 1.19–1.31). The risk increased with the number of MetS components including central obesity, dyslipidemia for high-density lipoprotein (HDL) cholesterol, hypertension, hyperglycemia, and dyslipidemia for triglycerides. Particularly for those with all five components (HR = 1.76, 95% CI = 1.51–2.04). Dyslipidemia for HDL cholesterol, hypertension, hyperglycemia, and dyslipidemia for triglycerides were independently associated with elevated dementia risk (p < 0.01). MetS was further linked to an increased risk of all-cause dementia (11%) and vascular dementia (VD, 50%) among individuals with SUA levels exceeding 400 μmol/L (all-cause dementia: HR = 1.11, 95% CI = 1.02–1.21; VD: HR = 1.50, 95% CI = 1.28–1.77). Conclusions Our study provides robust evidence supporting the association between MetS, its components, and dementia risk. These findings emphasize the importance of considering MetS and SUA levels in assessing dementia risk, offering valuable insights for prevention and management strategies.

Background and Objective: We explored dynamic changes in the choroid plexus (CP) in patients with relapsing-remitting multiple sclerosis (RRMS) and assessed its relationship with chronic lesion expansion and atrophy in various brain compartments. Methods: Fifty-seven RRMS patients were annually assessed for a minimum of 48 months with 3D FLAIR, pre- and post-contrast 3D T1 and diffusion-weighted magnetic resonance imaging (MRI). The CP was manually segmented at baseline and last follow-up. Results: The volume of CP significantly increased by 1.4% annually. However, the extent of CP enlargement varied considerably among individuals (ranging from -3.6 to 150.8 mm(3) or -0.2% to 6.3%). The magnitude of CP enlargement significantly correlated with central (r = 0.70, p < 0.001) and total brain atrophy (r = -0.57, p < 0.001), white (r = -0.61, p < 0.001) and deep grey matter atrophy (r = -0.60, p < 0.001). Progressive CP enlargement was significantly associated with the volume and extent of chronic lesion expansion (r = 0.60, p < 0.001), but not with the number or volume of new lesions. Conclusion: This study provides evidence of progressive CP enlargement in patients with RRMS. Our findings also demonstrate that enlargement of the CP volume is linked to the expansion of chronic lesions and neurodegeneration of periventricular white and grey matter in RRMS patients.

Background Magnetic resonance-guided focused ultrasound (MRgFUS) for treatment of essential tremor (ET) traditionally targets the ventral intermediate (Vim) nucleus. Recent strategies include a secondary lesion to the posterior subthalamic area (PSA). Objective The aim was to compare lesion characteristics, tremor improvement, and adverse events (AE) between patients in whom satisfactory tremor suppression was achieved with lesioning of the Vim alone and patients who required additional lesioning of the PSA. Methods Retrospective analysis of data collected from ET patients treated with MRgFUS at St Vincent's Hospital Sydney was performed. Clinical Rating Scale for Tremor (CRST), hand tremor score (HTS), and Quality of Life in Essential Tremor Questionnaire (QUEST) were collected pre- and posttreatment in addition to the prevalence of AEs. The lesion coordinates and overlap with the dentatorubrothalamic tract (DRTT) were evaluated using magnetic resonance imaging. Results Twenty-one patients were treated in Vim only, and 14 were treated with dual Vim-PSA lesions. Clinical data were available for 29 of the 35 patients (19 single target and 10 dual target). At follow-up (mean: 18.80 months) HTS, CRST, and QUEST in single-target patients improved by 57.97% (P < 0.001), 36.71% (P < 0.001), and 58.26% (P < 0.001), whereas dual-target patients improved by 68.34% (P < 0.001), 35.37% (P < 0.003), and 46.97% (P < 0.005), respectively. The Vim lesion of dual-target patients was further anterior relative to the posterior commissure (PC) (7.84 mm), compared with single-target patients (6.92 mm), with less DRTT involvement (14.85% vs. 23.21%). Dual-target patients exhibited a greater proportion of patients with acute motor AEs (100% vs. 58%); however, motor AE prevalence was similar in both groups at long-term follow-up (33% vs. 38%). Conclusion Posterior placement of lesions targeting the Vim may confer greater tremor suppression. The addition of a PSA lesion, in patients with inadequate tremor control despite Vim lesioning, had a trend toward better long-term tremor suppression; however, this approach was associated with greater prevalence of gait disturbance in the short term.

Training deep neural networks reliably requires access to large-scale datasets. However, obtaining such datasets can be challenging, especially in the context of neuroimaging analysis tasks, where the cost associated with image acquisition and annotation can be prohibitive. To mitigate both the time and financial costs associated with model development, a clear understanding of the amount of data required to train a satisfactory model is crucial. This paper focuses on an early stage phase of deep learning research, prior to model development, and proposes a strategic framework for estimating the amount of annotated data required to train patch-based segmentation networks. This framework includes the establishment of performance expectations using a novel Minor Boundary Adjustment for Threshold (MinBAT) method, and standardizing patch selection through the ROI-based Expanded Patch Selection (REPS) method. Our experiments demonstrate that tasks involving regions of interest (ROIs) with different sizes or shapes may yield variably acceptable Dice Similarity Coefficient (DSC) scores. By setting an acceptable DSC as the target, the required amount of training data can be estimated and even predicted as data accumulates. This approach could assist researchers and engineers in estimating the cost associated with data collection and annotation when defining a new segmentation task based on deep neural networks, ultimately contributing to their efficient translation to real-world applications.

The establishment of left–right asymmetry is a fundamental process in animal development. Interference with this process leads to a range of disorders collectively known as laterality defects, which manifest as abnormal arrangements of visceral organs. Among patients with laterality defects, congenital heart diseases (CHD) are prevalent. Through multiple model organisms, extant research has established that myosin-Id (MYO1D) deficiency causes laterality defects. This study investigated over a hundred cases and identified a novel biallelic variant of MYO1D (NM_015194: c.1531G>A; p.D511N) in a consanguineous family with complex CHD and laterality defects. Further examination of the proband revealed asthenoteratozoospermia and shortened sperm. Afterward, the effects of the D511N variant and another known MYO1D variant (NM_015194: c.2293C>T; p.P765S) were assessed. The assessment showed that both enhance the interaction with β-actin and SPAG6. Overall, this study revealed the genetic heterogeneity of this rare disease and found that MYO1D variants are correlated with laterality defects and CHD in humans. Furthermore, this research established a connection between sperm defects and MYO1D variants. It offers guidance for exploring infertility and reproductive health concerns. The findings provide a critical basis for advancing personalized medicine and genetic counseling.

Accurately measuring the evolution of Multiple Sclerosis (MS) with magnetic resonance imaging (MRI) critically informs understanding of disease progression and helps to direct therapeutic strategy. Deep learning models have shown promise for automatically segmenting MS lesions, but the scarcity of accurately annotated data hinders progress in this area. Obtaining sufficient data from a single clinical site is challenging and does not address the heterogeneous need for model robustness. Conversely, the collection of data from multiple sites introduces data privacy concerns and potential label noise due to varying annotation standards. To address this dilemma, we explore the use of the federated learning framework while considering label noise. Our approach enables collaboration among multiple clinical sites without compromising data privacy under a federated learning paradigm that incorporates a noise-robust training strategy based on label correction. Specifically, we introduce a Decoupled Hard Label Correction (DHLC) strategy that considers the imbalanced distribution and fuzzy boundaries of MS lesions, enabling the correction of false annotations based on prediction confidence. We also introduce a Centrally Enhanced Label Correction (CELC) strategy, which leverages the aggregated central model as a correction teacher for all sites, enhancing the reliability of the correction process. Extensive experiments conducted on two multi-site datasets demonstrate the effectiveness and robustness of our proposed methods, indicating their potential for clinical applications in multi-site collaborations to train better deep learning models with lower cost in data collection and annotation.

Background Low-intensity repetitive transcranial magnetic stimulation (rTMS), delivered as a daily intermittent theta burst stimulation (iTBS) for four consecutive weeks, increased the number of new oligodendrocytes in the adult mouse brain. Therefore, rTMS holds potential as a remyelinating intervention for people with multiple sclerosis (MS). Objective Primarily to determine the safety and tolerability of our rTMS protocol in people with MS. Secondary objectives include feasibility, blinding and an exploration of changes in magnetic resonance imaging (MRI) metrics, patient-reported outcome measures (PROMs) and cognitive or motor performance. Methods A randomised (2:1), placebo controlled, single blind, parallel group, phase 1 trial of 20 rTMS sessions (600 iTBS pulses per hemisphere; 25% maximum stimulator output), delivered over 4–5 weeks. Twenty participants were randomly assigned to ‘sham’ ( n = 7) or active rTMS ( n = 13), with the coil positioned at 90° or 0°, respectively. Results Five adverse events (AEs) including one serious AE reported. None were related to treatment. Protocol compliance was high (85%) and blinding successful. Within participant MRI metrics, PROMs and cognitive or motor performance were unchanged over time. Conclusion Twenty sessions of rTMS is safe and well tolerated in a small group of people with MS. The study protocol and procedures are feasible. Improvement of sham is warranted before further investigating safety and efficacy.