Evaluation of cervical spinal cord atrophy using a modified SIENA approach.

Spinal cord (SC) atrophy obtained from structural magnetic resonance imaging has gained relevance as an indicator of neurodegeneration in various neurological disorders. The common method to assess SC atrophy is by comparing numerical differences of the cross-sectional spinal cord area (CSA) between time points. However, this indirect approach leads to considerable variability in the obtained results. Studies showed that this limitation can be overcome by using a registration-based technique. The present study introduces the Structural Image Evaluation using Normalization of Atrophy on the Spinal Cord (SIENA-SC), which is an adapted version of the original SIENA method, designed to directly calculate the percentage of SC volume change over time from clinical brain MRI acquired with an extended field of view to cover the superior part of the cervical SC. In this work, we compared SIENA-SC with the Generalized Boundary Shift Integral (GBSI) and the CSA change. On a scan-rescan dataset, SIENA-SC was shown to have the lowest measurement error than the other two methods. When comparing a group of 190 Healthy Controls with a group of 65 Multiple Sclerosis patients, SIENA-SC provided significantly higher yearly rates of atrophy in patients than in controls and a lower sample size when measured for treatment effect sizes of 50%, 30% and 10%. Our findings indicate that SIENA-SC is a robust, reproducible, and sensitive approach for assessing longitudinal changes in spinal cord volume, providing neuroscientists with an accessible and automated tool able to reduce the need for manual intervention and minimize variability in measurements.

Improving criteria for dissemination in space in multiple sclerosis by including additional regions.

OBJECTIVE: We investigated the effects of adding regions to current dissemination in space (DIS) criteria for multiple sclerosis (MS). METHODS: Participants underwent brain, optic nerve, and spinal cord MRI. Baseline DIS was assessed by 2017 McDonald criteria and versions including optic nerve, temporal lobe, or corpus callosum as a fifth region (requiring 2/5), a version with all regions (requiring 3/7) and optic nerve variations requiring 3/5 and 4/5 regions. Performance was evaluated against MS diagnosis (2017 McDonald criteria) during follow-up. RESULTS: Eighty-four participants were recruited (53F, 32.8 ± 7.1 years). 2017 McDonald DIS criteria were 87% sensitive (95% CI: 76-94), 73% specific (50-89), and 83% accurate (74-91) in identifying MS. Modified criteria with optic nerve improved sensitivity to 98% (91-100), with specificity 33% (13-59) and accuracy 84% (74-91). Criteria including temporal lobe showed sensitivity 94% (84-98), specificity 50% (28-72), and accuracy 82% (72-90); criteria including corpus callosum showed sensitivity 90% (80-96), specificity 68% (45-86), and accuracy 85% (75-91). Criteria adding all three regions (3/7 required) had sensitivity 95% (87-99), specificity 55% (32-76), and accuracy 85% (75-91). When requiring 3/5 regions (optic nerve as the fifth), sensitivity was 82% (70-91), specificity 77% (55-92), and accuracy 81% (71-89); with 4/5 regions, sensitivity was 56% (43-69), specificity 95% (77-100), and accuracy 67% (56-77). INTERPRETATION: Optic nerve inclusion increased sensitivity while lowering specificity. Increasing required regions in optic nerve criteria increased specificity and decreased sensitivity. Results suggest considering the optic nerve for DIS. An option of 3/5 or 4/5 regions preserved specificity, and criteria adding all three regions had highest accuracy.

The evolving contribution of MRI measures towards the prediction of secondary progressive multiple sclerosis.

BACKGROUND: In multiple sclerosis (MS), both lesion accrual and brain atrophy predict clinical outcomes. However, it is unclear whether these prognostic features are equally relevant throughout the course of MS. Among 103 participants recruited following a clinically isolated syndrome (CIS) and followed up over 30 years, we explored (1) whether white matter lesions were prognostically more relevant earlier and brain atrophy later in the disease course towards development of secondary progressive (SP) disease; (2) if so, when the balance in prognostic contribution shifts and (3) whether optimised prognostic models predicting SP disease should include different features dependent on disease duration. METHODS: Binary logistic regression models were built using age, gender, brain lesion counts and locations, and linear atrophy measures (third ventricular width and medullary width) at each time point up to 20 years, using either single time point data alone or adjusted for baseline measures. RESULTS: By 30 years, 27 participants remained CIS while 60 had MS (26 SPMS and 16 MS-related death). Lesions counts were prognostically significant from baseline and at all later time points while linear atrophy measure models reached significance from 5 years. When adjusted for baseline, in combined MRI models including lesion count and linear atrophy measures, only lesion counts were significant predictors. In combined models including relapse measures, Expanded Disability Status Scale scores and MRI measures, only infratentorial lesions were significant predictors throughout. CONCLUSIONS: While SPMS progression is associated with brain atrophy, in predictive models only infratentorial lesions were consistently prognostically significant.

Finding the limits of deep learning clinical sensitivity with fractional anisotropy (FA) microstructure maps.

Background: Quantitative maps obtained with diffusion weighted (DW) imaging, such as fractional anisotropy (FA) -calculated by fitting the diffusion tensor (DT) model to the data,-are very useful to study neurological diseases. To fit this map accurately, acquisition times of the order of several minutes are needed because many noncollinear DW volumes must be acquired to reduce directional biases. Deep learning (DL) can be used to reduce acquisition times by reducing the number of DW volumes. We already developed a DL network named "one-minute FA," which uses 10 DW volumes to obtain FA maps, maintaining the same characteristics and clinical sensitivity of the FA maps calculated with the standard method using more volumes. Recent publications have indicated that it is possible to train DL networks and obtain FA maps even with 4 DW input volumes, far less than the minimum number of directions for the mathematical estimation of the DT. Methods: Here we investigated the impact of reducing the number of DW input volumes to 4 or 7, and evaluated the performance and clinical sensitivity of the corresponding DL networks trained to calculate FA, while comparing results also with those using our one-minute FA. Each network training was performed on the human connectome project open-access dataset that has a high resolution and many DW volumes, used to fit a ground truth FA. To evaluate the generalizability of each network, they were tested on two external clinical datasets, not seen during training, and acquired on different scanners with different protocols, as previously done. Results: Using 4 or 7 DW volumes, it was possible to train DL networks to obtain FA maps with the same range of values as ground truth - map, only when using HCP test data; pathological sensitivity was lost when tested using the external clinical datasets: indeed in both cases, no consistent differences were found between patient groups. On the contrary, our "one-minute FA" did not suffer from the same problem. Conclusion: When developing DL networks for reduced acquisition times, the ability to generalize and to generate quantitative biomarkers that provide clinical sensitivity must be addressed.

Grey Matter Atrophy and its Relationship with White Matter Lesions in Patients with Myelin Oligodendrocyte Glycoprotein Antibody-associated Disease, Aquaporin-4 Antibody-Positive Neuromyelitis Optica Spectrum Disorder, and Multiple Sclerosis.

OBJECTIVE: To evaluate: (1) the distribution of gray matter (GM) atrophy in myelin oligodendrocyte glycoprotein antibody-associated disease (MOGAD), aquaporin-4 antibody-positive neuromyelitis optica spectrum disorder (AQP4+NMOSD), and relapsing-remitting multiple sclerosis (RRMS); and (2) the relationship between GM volumes and white matter lesions in various brain regions within each disease. METHODS: A retrospective, multicenter analysis of magnetic resonance imaging data included patients with MOGAD/AQP4+NMOSD/RRMS in non-acute disease stage. Voxel-wise analyses and general linear models were used to evaluate the relevance of regional GM atrophy. For significant results (p < 0.05), volumes of atrophic areas are reported. RESULTS: We studied 135 MOGAD patients, 135 AQP4+NMOSD, 175 RRMS, and 144 healthy controls (HC). Compared with HC, MOGAD showed lower GM volumes in the temporal lobes, deep GM, insula, and cingulate cortex (75.79 cm3); AQP4+NMOSD in the occipital cortex (32.83 cm3); and RRMS diffusely in the GM (260.61 cm3). MOGAD showed more pronounced temporal cortex atrophy than RRMS (6.71 cm3), whereas AQP4+NMOSD displayed greater occipital cortex atrophy than RRMS (19.82 cm3). RRMS demonstrated more pronounced deep GM atrophy in comparison with MOGAD (27.90 cm3) and AQP4+NMOSD (47.04 cm3). In MOGAD, higher periventricular and cortical/juxtacortical lesions were linked to reduced temporal cortex, deep GM, and insula volumes. In RRMS, the diffuse GM atrophy was associated with lesions in all locations. AQP4+NMOSD showed no lesion/GM volume correlation. INTERPRETATION: GM atrophy is more widespread in RRMS compared with the other two conditions. MOGAD primarily affects the temporal cortex, whereas AQP4+NMOSD mainly involves the occipital cortex. In MOGAD and RRMS, lesion-related tract degeneration is associated with atrophy, but this link is absent in AQP4+NMOSD. ANN NEUROL 2024;96:276-288.

Improving explanation of motor disability with diffusion-based graph metrics at onset of the first demyelinating event.

BACKGROUND: Conventional magnetic resonance imaging (MRI) does not account for all disability in multiple sclerosis. OBJECTIVE: The objective was to assess the ability of graph metrics from diffusion-based structural connectomes to explain motor function beyond conventional MRI in early demyelinating clinically isolated syndrome (CIS). METHODS: A total of 73 people with CIS underwent conventional MRI, diffusion-weighted imaging and clinical assessment within 3 months from onset. A total of 28 healthy controls underwent MRI. Structural connectomes were produced. Differences between patients and controls were explored; clinical associations were assessed in patients. Linear regression models were compared to establish relevance of graph metrics over conventional MRI. RESULTS: Local efficiency (p = 0.045), clustering (p = 0.034) and transitivity (p = 0.036) were reduced in patients. Higher assortativity was associated with higher Expanded Disability Status Scale (EDSS) (ß = 74.9, p = 0.026) scores. Faster timed 25-foot walk (T25FW) was associated with higher assortativity (ß = 5.39, p = 0.026), local efficiency (ß = 27.1, p = 0.041) and clustering (ß = 36.1, p = 0.032) and lower small-worldness (ß = -3.27, p = 0.015). Adding graph metrics to conventional MRI improved EDSS (p = 0.045, ΔR2 = 4) and T25FW (p < 0.001, ΔR2 = 13.6) prediction. CONCLUSION: Graph metrics are relevant early in demyelination. They show differences between patients and controls and have relationships with clinical outcomes. Segregation (local efficiency, clustering, transitivity) was particularly relevant. Combining graph metrics with conventional MRI better explained disability.

Optic chiasm involvement in multiple sclerosis, aquaporin-4 antibody-positive neuromyelitis optica spectrum disorder and myelin oligodendrocyte glycoprotein-associated disease.

BACKGROUND: Optic neuritis (ON) is a common feature of inflammatory demyelinating diseases (IDDs) such as multiple sclerosis (MS), aquaporin 4-antibody neuromyelitis optica spectrum disorder (AQP4 + NMOSD) and myelin oligodendrocyte glycoprotein antibody-associated disease (MOGAD). However, the involvement of the optic chiasm (OC) in IDD has not been fully investigated. AIMS: To examine OC differences in non-acute IDD patients with (ON+) and without ON (ON-) using magnetisation transfer ratio (MTR), to compare differences between MS, AQP4 + NMOSD and MOGAD and understand their associations with other neuro-ophthalmological markers. METHODS: Twenty-eight relapsing-remitting multiple sclerosis (RRMS), 24 AQP4 + NMOSD, 28 MOGAD patients and 32 healthy controls (HCs) underwent clinical evaluation, MRI and optical coherence tomography (OCT) scan. Multivariable linear regression models were applied. RESULTS: ON + IDD patients showed lower OC MTR than HCs (28.87 ± 4.58 vs 31.65 ± 4.93; p = 0.004). When compared with HCs, lower OC MTR was found in ON + AQP4 + NMOSD (28.55 ± 4.18 vs 31.65 ± 4.93; p = 0.020) and MOGAD (28.73 ± 4.99 vs 31.65 ± 4.93; p = 0.007) and in ON- AQP4 + NMOSD (28.37 ± 7.27 vs 31.65 ± 4.93; p = 0.035). ON+ RRMS had lower MTR than ON- RRMS (28.87 ± 4.58 vs 30.99 ± 4.76; p = 0.038). Lower OC MTR was associated with higher number of ON (regression coefficient (RC) = -1.15, 95% confidence interval (CI) = -1.819 to -0.490, p = 0.001), worse visual acuity (RC = -0.026, 95% CI = -0.041 to -0.011, p = 0.001) and lower peripapillary retinal nerve fibre layer (pRNFL) thickness (RC = 1.129, 95% CI = 0.199 to 2.059, p = 0.018) when considering the whole IDD group. CONCLUSION: OC microstructural damage indicates prior ON in IDD and is linked to reduced vision and thinner pRNFL.

Generalizable disease detection using model ensemble on chest X-ray images.

In the realm of healthcare, the demand for swift and precise diagnostic tools has been steadily increasing. This study delves into a comprehensive performance analysis of three pre-trained convolutional neural network (CNN) architectures: ResNet50, DenseNet121, and Inception-ResNet-v2. To ensure the broad applicability of our approach, we curated a large-scale dataset comprising a diverse collection of chest X-ray images, that included both positive and negative cases of COVID-19. The models' performance was evaluated using separate datasets for internal validation (from the same source as the training images) and external validation (from different sources). Our examination uncovered a significant drop in network efficacy, registering a 10.66% reduction for ResNet50, a 36.33% decline for DenseNet121, and a 19.55% decrease for Inception-ResNet-v2 in terms of accuracy. Best results were obtained with DenseNet121 achieving the highest accuracy at 96.71% in internal validation and Inception-ResNet-v2 attaining 76.70% accuracy in external validation. Furthermore, we introduced a model ensemble approach aimed at improving network performance when making inferences on images from diverse sources beyond their training data. The proposed method uses uncertainty-based weighting by calculating the entropy in order to assign appropriate weights to the outputs of each network. Our results showcase the effectiveness of the ensemble method in enhancing accuracy up to 97.38% for internal validation and 81.18% for external validation, while maintaining a balanced ability to detect both positive and negative cases.

What contributes to disability in progressive MS? A brain and cervical cord-matched quantitative MRI study.

BACKGROUND: We assessed the ability of a brain-and-cord-matched quantitative magnetic resonance imaging (qMRI) protocol to differentiate patients with progressive multiple sclerosis (PMS) from controls, in terms of normal-appearing (NA) tissue abnormalities, and explain disability. METHODS: A total of 27 patients and 16 controls were assessed on the Expanded Disability Status Scale (EDSS), 25-foot timed walk (TWT), 9-hole peg (9HPT) and symbol digit modalities (SDMT) tests. All underwent 3T brain and (C2-C3) cord structural imaging and qMRI (relaxometry, quantitative magnetisation transfer, multi-shell diffusion-weighted imaging), using a fast brain-and-cord-matched protocol with brain-and-cord-unified imaging readouts. Lesion and NA-tissue volumes and qMRI metrics reflecting demyelination and axonal loss were obtained. Random forest analyses identified the most relevant volumetric/qMRI measures to clinical outcomes. Confounder-adjusted linear regression estimated the actual MRI-clinical associations. RESULTS: Several qMRI/volumetric differences between patients and controls were observed (p < 0.01). Higher NA-deep grey matter quantitative-T1 (EDSS: beta = 7.96, p = 0.006; 9HPT: beta = -0.09, p = 0.004), higher NA-white matter orientation dispersion index (TWT: beta = -3.21, p = 0.005; SDMT: beta = -847.10, p < 0.001), lower whole-cord bound pool fraction (9HPT: beta = 0.79, p = 0.001) and higher NA-cortical grey matter quantitative-T1 (SDMT = -94.31, p < 0.001) emerged as particularly relevant predictors of greater disability. CONCLUSION: Fast brain-and-cord-matched qMRI protocols are feasible and identify demyelination - combined with other mechanisms - as key for disability accumulation in PMS.

Role of the repeat expansion size in predicting age of onset and severity in RFC1 disease.

RFC1 disease, caused by biallelic repeat expansion in RFC1, is clinically heterogeneous in terms of age of onset, disease progression and phenotype. We investigated the role of the repeat size in influencing clinical variables in RFC1 disease. We also assessed the presence and role of meiotic and somatic instability of the repeat. In this study, we identified 553 patients carrying biallelic RFC1 expansions and measured the repeat expansion size in 392 cases. Pearson's coefficient was calculated to assess the correlation between the repeat size and age at disease onset. A Cox model with robust cluster standard errors was adopted to describe the effect of repeat size on age at disease onset, on age at onset of each individual symptoms, and on disease progression. A quasi-Poisson regression model was used to analyse the relationship between phenotype and repeat size. We performed multivariate linear regression to assess the association of the repeat size with the degree of cerebellar atrophy. Meiotic stability was assessed by Southern blotting on first-degree relatives of 27 probands. Finally, somatic instability was investigated by optical genome mapping on cerebellar and frontal cortex and unaffected peripheral tissue from four post-mortem cases. A larger repeat size of both smaller and larger allele was associated with an earlier age at neurological onset [smaller allele hazard ratio (HR) = 2.06, P < 0.001; larger allele HR = 1.53, P < 0.001] and with a higher hazard of developing disabling symptoms, such as dysarthria or dysphagia (smaller allele HR = 3.40, P < 0.001; larger allele HR = 1.71, P = 0.002) or loss of independent walking (smaller allele HR = 2.78, P < 0.001; larger allele HR = 1.60; P < 0.001) earlier in disease course. Patients with more complex phenotypes carried larger expansions [smaller allele: complex neuropathy rate ratio (RR) = 1.30, P = 0.003; cerebellar ataxia, neuropathy and vestibular areflexia syndrome (CANVAS) RR = 1.34, P < 0.001; larger allele: complex neuropathy RR = 1.33, P = 0.008; CANVAS RR = 1.31, P = 0.009]. Furthermore, larger repeat expansions in the smaller allele were associated with more pronounced cerebellar vermis atrophy (lobules I-V ß = -1.06, P < 0.001; lobules VI-VII ß = -0.34, P = 0.005). The repeat did not show significant instability during vertical transmission and across different tissues and brain regions. RFC1 repeat size, particularly of the smaller allele, is one of the determinants of variability in RFC1 disease and represents a key prognostic factor to predict disease onset, phenotype and severity. Assessing the repeat size is warranted as part of the diagnostic test for RFC1 expansion.

Associations between cortical lesions, optic nerve damage, and disability at the onset of multiple sclerosis: insights into neurodegenerative processes.

BACKGROUND: Multiple sclerosis cortical lesions are areas of demyelination and neuroaxonal loss. Retinal layer thickness, measured with optical coherence tomography (OCT), is an emerging biomarker of neuroaxonal loss. Studies have reported correlations between cortical lesions and retinal layer thinning in established multiple sclerosis, suggesting a shared pathophysiological process. Here, we assessed the correlation between cortical lesions and OCT metrics at the onset of multiple sclerosis, examining, for the first time, associations with physical or cognitive disability. OBJECTIVE: To examine the relationship between cortical lesions, optic nerve and retinal layer thicknesses, and physical and cognitive disability at the first demyelinating event. METHODS: Thirty-nine patients and 22 controls underwent 3T-MRI, optical coherence tomography, and clinical tests. We identified cortical lesions on phase-sensitive inversion recovery sequences, including occipital cortex lesions. We measured the estimated total intracranial volume and the white matter lesion volume. OCT metrics included peripapillary retinal nerve fibre layer (pRNFL), ganglion cell and inner plexiform layer (GCIPL) and inner nuclear layer (INL) thicknesses. RESULTS: Higher total cortical and leukocortical lesion volumes correlated with thinner pRNFL (B = -0.0005, 95 % CI -0.0008 to -0.0001, p = 0.01; B = -0.0005, 95 % CI -0.0008 to -0.0001, p = 0.01, respectively). Leukocortical lesion number correlated with colour vision deficits (B = 0.58, 95 %CI 0.039 to 1,11, p = 0.036). Thinner GCIPL correlated with a higher Expanded Disability Status Scale (B = -0.06, 95 % CI -1.1 to -0.008, p = 0.026). MS diagnosis (n = 18) correlated with higher cortical and leukocortical lesion numbers (p = 0.004 and p = 0.003), thinner GCIPL (p = 0.029) and INL (p = 0.041). CONCLUSION: The association between cortical lesions and axonal damage in the optic nerve reinforces the role of neurodegenerative processes in MS pathogenesis at onset.

Visual Snow Syndrome Improves With Modulation of Resting-State Functional MRI Connectivity After Mindfulness-Based Cognitive Therapy: An Open-Label Feasibility Study.

BACKGROUND: Visual snow syndrome (VSS) is associated with functional connectivity (FC) dysregulation of visual networks (VNs). We hypothesized that mindfulness-based cognitive therapy, customized for visual symptoms (MBCT-vision), can treat VSS and modulate dysfunctional VNs. METHODS: An open-label feasibility study for an 8-week MBCT-vision treatment program was conducted. Primary (symptom severity; impact on daily life) and secondary (WHO-5; CORE-10) outcomes at Week 9 and Week 20 were compared with baseline. Secondary MRI outcomes in a subcohort compared resting-state functional and diffusion MRI between baseline and Week 20. RESULTS: Twenty-one participants (14 male participants, median 30 years, range 22-56 years) recruited from January 2020 to October 2021. Two (9.5%) dropped out. Self-rated symptom severity (0-10) improved: baseline (median [interquartile range (IQR)] 7 [6-8]) vs Week 9 (5.5 [3-7], P = 0.015) and Week 20 (4 [3-6], P < 0.001), respectively. Self-rated impact of symptoms on daily life (0-10) improved: baseline (6 [5-8]) vs Week 9 (4 [2-5], P = 0.003) and Week 20 (2 [1-3], P < 0.001), respectively. WHO-5 Wellbeing (0-100) improved: baseline (median [IQR] 52 [36-56]) vs Week 9 (median 64 [47-80], P = 0.001) and Week 20 (68 [48-76], P < 0.001), respectively. CORE-10 Distress (0-40) improved: baseline (15 [12-20]) vs Week 9 (12.5 [11-16.5], P = 0.003) and Week 20 (11 [10-14], P = 0.003), respectively. Within-subject fMRI analysis found reductions between baseline and Week 20, within VN-related FC in the i) left lateral occipital cortex (size = 82 mL, familywise error [FWE]-corrected P value = 0.006) and ii) left cerebellar lobules VIIb/VIII (size = 65 mL, FWE-corrected P value = 0.02), and increases within VN-related FC in the precuneus/posterior cingulate cortex (size = 69 mL, cluster-level FWE-corrected P value = 0.02). CONCLUSIONS: MBCT-vision was a feasible treatment for VSS, improved symptoms and modulated FC of VNs. This study also showed proof-of-concept for intensive mindfulness interventions in the treatment of neurological conditions.

Treatment reduces the incidence of newly appearing multiple sclerosis lesions evolving into chronic active, slowly expanding lesions: A retrospective analysis.

BACKGROUND AND PURPOSE: Newly appearing lesions in multiple sclerosis (MS) may evolve into chronically active, slowly expanding lesions (SELs), leading to sustained disability progression. The aim of this study was to evaluate the incidence of newly appearing lesions developing into SELs, and their correlation to clinical evolution and treatment. METHODS: A retrospective analysis of a fingolimod trial in primary progressive MS (PPMS; INFORMS, NCT00731692) was undertaken. Data were available from 324 patients with magnetic resonance imaging scans up to 3 years after screening. New lesions at year 1 were identified with convolutional neural networks, and SELs obtained through a deformation-based method. Clinical disability was assessed annually by Expanded Disability Status Scale (EDSS), Nine-Hole Peg Test, Timed 25-Foot Walk, and Paced Auditory Serial Addition Test. Linear, logistic, and mixed-effect models were used to assess the relationship between the Jacobian expansion in new lesions and SELs, disability scores, and treatment status. RESULTS: One hundred seventy patients had ≥1 new lesions at year 1 and had a higher lesion count at screening compared to patients with no new lesions (median = 27 vs. 22, p = 0.007). Among the new lesions (median = 2 per patient), 37% evolved into definite or possible SELs. Higher SEL volume and count were associated with EDSS worsening and confirmed disability progression. Treated patients had lower volume and count of definite SELs (ß = -0.04, 95% confidence interval [CI] = -0.07 to -0.01, p = 0.015; ß = -0.36, 95% CI = -0.67 to -0.06, p = 0.019, respectively). CONCLUSIONS: Incident chronic active lesions are common in PPMS, and fingolimod treatment can reduce their number.

Multi-class glioma segmentation on real-world data with missing MRI sequences: comparison of three deep learning algorithms.

This study tests the generalisability of three Brain Tumor Segmentation (BraTS) challenge models using a multi-center dataset of varying image quality and incomplete MRI datasets. In this retrospective study, DeepMedic, no-new-Unet (nn-Unet), and NVIDIA-net (nv-Net) were trained and tested using manual segmentations from preoperative MRI of glioblastoma (GBM) and low-grade gliomas (LGG) from the BraTS 2021 dataset (1251 in total), in addition to 275 GBM and 205 LGG acquired clinically across 12 hospitals worldwide. Data was split into 80% training, 5% validation, and 15% internal test data. An additional external test-set of 158 GBM and 69 LGG was used to assess generalisability to other hospitals' data. All models' median Dice similarity coefficient (DSC) for both test sets were within, or higher than, previously reported human inter-rater agreement (range of 0.74-0.85). For both test sets, nn-Unet achieved the highest DSC (internal = 0.86, external = 0.93) and the lowest Hausdorff distances (10.07, 13.87 mm, respectively) for all tumor classes (p < 0.001). By applying Sparsified training, missing MRI sequences did not statistically affect the performance. nn-Unet achieves accurate segmentations in clinical settings even in the presence of incomplete MRI datasets. This facilitates future clinical adoption of automated glioma segmentation, which could help inform treatment planning and glioma monitoring.

Unsupervised Cluster Analysis Reveals Distinct Subtypes of ME/CFS Patients Based on Peak Oxygen Consumption and SF-36 Scores.

PURPOSE: Myalgic encephalomyelitis, commonly referred to as chronic fatigue syndrome (ME/CFS), is a severe, disabling chronic disease and an objective assessment of prognosis is crucial to evaluate the efficacy of future drugs. Attempts are ongoing to find a biomarker to objectively assess the health status of (ME/CFS), patients. This study therefore aims to demonstrate that oxygen consumption is a biomarker of ME/CFS provides a method to classify patients diagnosed with ME/CFS based on their responses to the Short Form-36 (SF-36) questionnaire, which can predict oxygen consumption using cardiopulmonary exercise testing (CPET). METHODS: Two datasets were used in the study. The first contained SF-36 responses from 2,347 validated records of ME/CFS diagnosed participants, and an unsupervised machine learning model was developed to cluster the data. The second dataset was used as a validation set and included the cardiopulmonary exercise test (CPET) results of 239 participants diagnosed with ME/CFS. Participants from this dataset were grouped by peak oxygen consumption according to Weber's classification. The SF-36 questionnaire was correctly completed by only 92 patients, who were clustered using the machine learning model. Two categorical variables were then entered into a contingency table: the cluster with values {0,1} and Weber classification {A, B, C, D} were assigned. Finally, the Chi-square test of independence was used to assess the statistical significance of the relationship between the two parameters. FINDINGS: The results indicate that the Weber classification is directly linked to the score on the SF-36 questionnaire. Furthermore, the 36-response matrix in the machine learning model was shown to give more reliable results than the subscale matrix (p - value < 0.05) for classifying patients with ME/CFS. IMPLICATIONS: Low oxygen consumption on CPET can be considered a biomarker in patients with ME/CFS. Our analysis showed a close relationship between the cluster based on their SF-36 questionnaire score and the Weber classification, which was based on peak oxygen consumption during CPET. The dataset for the training model comprised raw responses from the SF-36 questionnaire, which is proven to better preserve the original information, thus improving the quality of the model.

A synthetic data generation system for myalgic encephalomyelitis/chronic fatigue syndrome questionnaires.

Artificial intelligence or machine-learning-based models have proven useful for better understanding various diseases in all areas of health science. Myalgic Encephalomyelitis or chronic fatigue syndrome (ME/CFS) lacks objective diagnostic tests. Some validated questionnaires are used for diagnosis and assessment of disease progression. The availability of a sufficiently large database of these questionnaires facilitates research into new models that can predict profiles that help to understand the etiology of the disease. A synthetic data generator provides the scientific community with databases that preserve the statistical properties of the original, free of legal restrictions, for use in research and education. The initial databases came from the Vall Hebron Hospital Specialized Unit in Barcelona, Spain. 2522 patients diagnosed with ME/CFS were analyzed. Their answers to questionnaires related to the symptoms of this complex disease were used as training datasets. They have been fed for deep learning algorithms that provide models with high accuracy [0.69-0.81]. The final model requires SF-36 responses and returns responses from HAD, SCL-90R, FIS8, FIS40, and PSQI questionnaires. A highly reliable and easy-to-use synthetic data generator is offered for research and educational use in this disease, for which there is currently no approved treatment.

Networks of microstructural damage predict disability in multiple sclerosis.

BACKGROUND: Network-based measures are emerging MRI markers in multiple sclerosis (MS). We aimed to identify networks of white (WM) and grey matter (GM) damage that predict disability progression and cognitive worsening using data-driven methods. METHODS: We analysed data from 1836 participants with different MS phenotypes (843 in a discovery cohort and 842 in a replication cohort). We calculated standardised T1-weighted/T2-weighted (sT1w/T2w) ratio maps in brain GM and WM, and applied spatial independent component analysis to identify networks of covarying microstructural damage. Clinical outcomes were Expanded Disability Status Scale worsening confirmed at 24 weeks (24-week confirmed disability progression (CDP)) and time to cognitive worsening assessed by the Symbol Digit Modalities Test (SDMT). We used Cox proportional hazard models to calculate predictive value of network measures. RESULTS: We identified 8 WM and 7 GM sT1w/T2w networks (of regional covariation in sT1w/T2w measures) in both cohorts. Network loading represents the degree of covariation in regional T1/T2 ratio within a given network. The loading factor in the anterior corona radiata and temporo-parieto-frontal components were associated with higher risks of developing CDP both in the discovery (HR=0.85, p<0.05 and HR=0.83, p<0.05, respectively) and replication cohorts (HR=0.84, p<0.05 and HR=0.80, p<0.005, respectively). The decreasing or increasing loading factor in the arcuate fasciculus, corpus callosum, deep GM, cortico-cerebellar patterns and lesion load were associated with a higher risk of developing SDMT worsening both in the discovery (HR=0.82, p<0.01; HR=0.87, p<0.05; HR=0.75, p<0.001; HR=0.86, p<0.05 and HR=1.27, p<0.0001) and replication cohorts (HR=0.82, p<0.005; HR=0.73, p<0.0001; HR=0.80, p<0.005; HR=0.85, p<0.01 and HR=1.26, p<0.0001). CONCLUSIONS: GM and WM networks of microstructural changes predict disability and cognitive worsening in MS. Our approach may be used to identify patients at greater risk of disability worsening and stratify cohorts in treatment trials.

Cardiovascular risk factors in secondary progressive multiple sclerosis: A cross-sectional analysis from the MS-STAT2 randomized controlled trial.

BACKGROUND AND PURPOSE: There is increasing evidence that cardiovascular risk (CVR) contributes to disability progression in multiple sclerosis (MS). CVR is particularly prevalent in secondary progressive MS (SPMS) and can be quantified through validated composite CVR scores. The aim was to examine the cross-sectional relationships between excess modifiable CVR, whole and regional brain atrophy on magnetic resonance imaging, and disability in patients with SPMS. METHODS: Participants had SPMS, and data were collected at enrolment into the MS-STAT2 trial. Composite CVR scores were calculated using the QRISK3 software. Prematurely achieved CVR due to modifiable risk factors was expressed as QRISK3 premature CVR, derived through reference to the normative QRISK3 dataset and expressed in years. Associations were determined with multiple linear regressions. RESULTS: For the 218 participants, mean age was 54 years and median Expanded Disability Status Scale was 6.0. Each additional year of prematurely achieved CVR was associated with a 2.7 mL (beta coefficient; 95% confidence interval 0.8-4.7; p = 0.006) smaller normalized whole brain volume. The strongest relationship was seen for the cortical grey matter (beta coefficient 1.6 mL per year; 95% confidence interval 0.5-2.7; p = 0.003), and associations were also found with poorer verbal working memory performance. Body mass index demonstrated the strongest relationships with normalized brain volumes, whilst serum lipid ratios demonstrated strong relationships with verbal and visuospatial working memory performance. CONCLUSIONS: Prematurely achieved CVR is associated with lower normalized brain volumes in SPMS. Future longitudinal analyses of this clinical trial dataset will be important to determine whether CVR predicts future disease worsening.

Diffusion-based structural connectivity patterns of multiple sclerosis phenotypes.

BACKGROUND: We aimed to describe the severity of the changes in brain diffusion-based connectivity as multiple sclerosis (MS) progresses and the microstructural characteristics of these networks that are associated with distinct MS phenotypes. METHODS: Clinical information and brain MRIs were collected from 221 healthy individuals and 823 people with MS at 8 MAGNIMS centres. The patients were divided into four clinical phenotypes: clinically isolated syndrome, relapsing-remitting, secondary progressive and primary progressive. Advanced tractography methods were used to obtain connectivity matrices. Then, differences in whole-brain and nodal graph-derived measures, and in the fractional anisotropy of connections between groups were analysed. Support vector machine algorithms were used to classify groups. RESULTS: Clinically isolated syndrome and relapsing-remitting patients shared similar network changes relative to controls. However, most global and local network properties differed in secondary progressive patients compared with the other groups, with lower fractional anisotropy in most connections. Primary progressive participants had fewer differences in global and local graph measures compared with clinically isolated syndrome and relapsing-remitting patients, and reductions in fractional anisotropy were only evident for a few connections. The accuracy of support vector machine to discriminate patients from healthy controls based on connection was 81%, and ranged between 64% and 74% in distinguishing among the clinical phenotypes. CONCLUSIONS: In conclusion, brain connectivity is disrupted in MS and has differential patterns according to the phenotype. Secondary progressive is associated with more widespread changes in connectivity. Additionally, classification tasks can distinguish between MS types, with subcortical connections being the most important factor.

Patterns of inflammation, microstructural alterations, and sodium accumulation define multiple sclerosis subtypes after 15 years from onset.

Introduction: Conventional MRI is routinely used for the characterization of pathological changes in multiple sclerosis (MS), but due to its lack of specificity is unable to provide accurate prognoses, explain disease heterogeneity and reconcile the gap between observed clinical symptoms and radiological evidence. Quantitative MRI provides measures of physiological abnormalities, otherwise invisible to conventional MRI, that correlate with MS severity. Analyzing quantitative MRI measures through machine learning techniques has been shown to improve the understanding of the underlying disease by better delineating its alteration patterns. Methods: In this retrospective study, a cohort of healthy controls (HC) and MS patients with different subtypes, followed up 15 years from clinically isolated syndrome (CIS), was analyzed to produce a multi-modal set of quantitative MRI features encompassing relaxometry, microstructure, sodium ion concentration, and tissue volumetry. Random forest classifiers were used to train a model able to discriminate between HC, CIS, relapsing remitting (RR) and secondary progressive (SP) MS patients based on these features and, for each classification task, to identify the relative contribution of each MRI-derived tissue property to the classification task itself. Results and discussion: Average classification accuracy scores of 99 and 95% were obtained when discriminating HC and CIS vs. SP, respectively; 82 and 83% for HC and CIS vs. RR; 76% for RR vs. SP, and 79% for HC vs. CIS. Different patterns of alterations were observed for each classification task, offering key insights in the understanding of MS phenotypes pathophysiology: atrophy and relaxometry emerged particularly in the classification of HC and CIS vs. MS, relaxometry within lesions in RR vs. SP, sodium ion concentration in HC vs. CIS, and microstructural alterations were involved across all tasks.