Trans-Synaptic Degeneration in the Visual Pathway in Patients With Myelin Oligodendrocyte Glycoprotein Antibody-Associated Disease.

BACKGROUND AND OBJECTIVES: We aimed to assess the presence of retinal neurodegeneration independent of optic neuritis (ON) in myelin oligodendrocyte glycoprotein antibody-associated disease (MOGAD) and to investigate the development of trans-synaptic anterograde degeneration in these patients after ON. METHODS: Cross-sectional, retrospective study of 34 adult patients with MOGAD and 23 healthy controls (HC). Clinical, optical coherence tomography (OCT), and MRI data were collected. Peripapillary retinal nerve fiber layer (pRNFL) and ganglion cell inner plexiform layer (GCIPL) were obtained using Heidelberg Spectralis. FreeSurfer7 was used to obtain the lateral geniculate nucleus (LGN), occipital volume fractions (to total estimated intracranial volume), and occipital cortical thickness. For the anterior visual pathway, the analysis was conducted using eyes, classified based on the history of ON (Eye-ON and Eye-NON) and compared with Eye-HC. The analysis of OCT and brain volumetric measures was conducted comparing MOGAD-ON, MOGAD-NON, and HC groups. The analysis of covariance with a Bonferroni-adjusted post hoc test was used to test differences between groups and linear regression analysis to evaluate OCT/MRI associations; age and sex were considered as covariates. RESULTS: 24 (70.5%) patients had a prior ON. Median pRNFL and GCIPL thickness (um) was significantly reduced in Eye-ON vs EyeNON and HC (pRNFL: 69.4 (17.3), 89.6 (13.7), 98.2 (11.7), p < 0.001; GCIPL: 55.8 (8.7), 67.39 (8.7), 72.6 (4.5), p < 0.001). pRNFL and GCIPL thickness had a negative correlation with the number of ON episodes (p = 0.025 and p = 0.031, respectively). LGN volume fraction was significantly lower in patients with MOGAD-ON than in HC (0.33 (0.05) vs 0.39 (0.04), p = 0.002). The occipital cortical thickness was lower in MOGAD-ON compared with MOGAD-NON and HC (p = 0.010). In patients with MOGAD-ON, pRNFL correlated with LGN volume (p = 0.006), occipital thickness (p = 0.002), and the medial occipital cortex (p = 0.002), but not the lateral occipital lobe. DISCUSSION: Compared with HC, MOGAD-ON exhibits reduced retinal thickness, primarily influenced by the presence and the number of prior ON episodes. Moreover, MOGAD-ON demonstrates significant atrophy in the retinal, subcortical, and cortical regions of the visual pathway, distinguishing them from MOGAD-NON and HC. These findings suggest that in patients with MOGAD neurodegeneration is tightly correlated with damage to the involved pathway.

Optic chiasm manual and automated measurements in sub-acute optic neuritis with OCT and MRI correlations.

PURPOSE: The optic chiasm (OC) is a central structure in the visual pathway and can be visualized in conventional MRI, but no consensus regarding its measurement has been defined. We aim to investigate the most reproducible manual approach to OC measurement and to explore associations of OC with optical coherence tomography (OCT) parameters, and automatic brain segmentation (FreeSurfer) in subacute optic neuritis (sON), multiple sclerosis without optic neuritis (MSwoON), and healthy subjects (HS). MATERIALS AND METHODS: We reproduced two previously reported methodologies and implemented a new proposed simplified approach, entitled optic chiasm mean area (OCMA). The intra and inter-rater reliability and reproducibility were assessed through the intraclass correlation (ICC) and Dice similarity (DSC) coefficients. Partial correlations were calculated to gauge the associations between OCMA fraction (OCMA divided by total intracranial volume), brain regional segmentations derived from FreeSurfer, and OCT parameters. RESULTS: We have analysed 43 sON, 20 MSwoON, and 20 HS. OCMA presented better results for reliability in both intra- and inter-rater analysis (excellent ICC and DSC with over 80% overlap between masks), as compared to the other two approaches. OCMA fraction was associated with OC volume fraction obtained with Freesurfer in all groups, brain parenchymal fraction, and OCT parameters in MSwoON. CONCLUSIONS: The OCMA is a simplified approach to measure OC atrophy, has a higher reliability than the current approaches and shows association with an automated method. OC-derived measures seem to reflect diffuse neurodegenerative damage, whereas, in patients with subacute ON, it may be associated with local damage.

Global and Regional Deep Learning Models for Multiple Sclerosis Stratification From MRI.

BACKGROUND: The combination of anatomical MRI and deep learning-based methods such as convolutional neural networks (CNNs) is a promising strategy to build predictive models of multiple sclerosis (MS) prognosis. However, studies assessing the effect of different input strategies on model's performance are lacking. PURPOSE: To compare whole-brain input sampling strategies and regional/specific-tissue strategies, which focus on a priori known relevant areas for disability accrual, to stratify MS patients based on their disability level. STUDY TYPE: Retrospective. SUBJECTS: Three hundred nineteen MS patients (382 brain MRI scans) with clinical assessment of disability level performed within the following 6 months (~70% training/~15% validation/~15% inference in-house dataset) and 440 MS patients from multiple centers (independent external validation cohort). FIELD STRENGTH/SEQUENCE: Single vendor 1.5 T or 3.0 T. Magnetization-Prepared Rapid Gradient-Echo and Fluid-Attenuated Inversion Recovery sequences. ASSESSMENT: A 7-fold patient cross validation strategy was used to train a 3D-CNN to classify patients into two groups, Expanded Disability Status Scale score (EDSS) ≥ 3.0 or EDSS < 3.0. Two strategies were investigated: 1) a global approach, taking the whole brain volume as input and 2) regional approaches using five different regions-of-interest: white matter, gray matter, subcortical gray matter, ventricles, and brainstem structures. The performance of the models was assessed in the in-house and the independent external cohorts. STATISTICAL TESTS: Balanced accuracy, sensitivity, specificity, area under receiver operating characteristic (ROC) curve (AUC). RESULTS: With the in-house dataset, the gray matter regional model showed the highest stratification accuracy (81%), followed by the global approach (79%). In the external dataset, without any further retraining, an accuracy of 72% was achieved for the white matter model and 71% for the global approach. DATA CONCLUSION: The global approach offered the best trade-off between internal performance and external validation to stratify MS patients based on accumulated disability. EVIDENCE LEVEL: 4 TECHNICAL EFFICACY: Stage 2.

Deciphering multiple sclerosis disability with deep learning attention maps on clinical MRI.

The application of convolutional neural networks (CNNs) to MRI data has emerged as a promising approach to achieving unprecedented levels of accuracy when predicting the course of neurological conditions, including multiple sclerosis, by means of extracting image features not detectable through conventional methods. Additionally, the study of CNN-derived attention maps, which indicate the most relevant anatomical features for CNN-based decisions, has the potential to uncover key disease mechanisms leading to disability accumulation. From a cohort of patients prospectively followed up after a first demyelinating attack, we selected those with T1-weighted and T2-FLAIR brain MRI sequences available for image analysis and a clinical assessment performed within the following six months (N = 319). Patients were divided into two groups according to expanded disability status scale (EDSS) score: ≥3.0 and < 3.0. A 3D-CNN model predicted the class using whole-brain MRI scans as input. A comparison with a logistic regression (LR) model using volumetric measurements as explanatory variables and a validation of the CNN model on an independent dataset with similar characteristics (N = 440) were also performed. The layer-wise relevance propagation method was used to obtain individual attention maps. The CNN model achieved a mean accuracy of 79% and proved to be superior to the equivalent LR-model (77%). Additionally, the model was successfully validated in the independent external cohort without any re-training (accuracy = 71%). Attention-map analyses revealed the predominant role of frontotemporal cortex and cerebellum for CNN decisions, suggesting that the mechanisms leading to disability accrual exceed the mere presence of brain lesions or atrophy and probably involve how damage is distributed in the central nervous system.

Spinal cord reserve in multiple sclerosis.

BACKGROUND: The spinal cord (SC) is a preferential target of multiple sclerosis (MS) damage highly relevant towards disability. Differential impact of such damage could be due to the initial amount of SC tissue, as described for the brain parenchyma (brain reserve concept). We aimed to test the existence of SC reserve by using spinal canal area (SCaA) as a proxy. METHODS: Brain sagittal three-dimensional T1-weighted scans covering down to C5 level were acquired in 2930 people with MS and 43 healthy controls (HCs) in a cross-sectional, multicentre study. SC area (SCA) and SCaA were obtained with the Spinal Cord Toolbox. Demographical data and patient-derived disability scores were obtained. SC parameters were compared between groups with age-adjusted and sex-adjusted linear regression models. The main outcome of the study, the existence of an association between SCaA and Patient Determined Disease Steps, was tested with scaled linear models. RESULTS: 1747 persons with MS (mean age: 46.35 years; 73.2% female) and 42 HCs (mean age: 45.56 years; 78.6% female) were analysed after exclusion of post-processing errors and application of quality criteria. SCA (60.41 mm2 vs 65.02 mm2, p<0.001) was lower in people with MS compared with HC; no differences in SCaA were observed (213.24 mm2 vs 212.61 mm2, p=0.125). Adjusted scaled linear models showed that a larger SCaA was significantly associated with lower scores on Patient Determined Disease Steps (beta coefficient: -0.12, p=0.0124) independently of spinal cord atrophy, brain T2 lesion volume, age and sex. CONCLUSIONS: A larger SCaA may be protective against disability in MS, possibly supporting the existence of SC reserve.

Relationship between paramagnetic rim lesions and slowly expanding lesions in multiple sclerosis.

BACKGROUND: Magnetic resonance imaging (MRI) markers for chronic active lesions in MS include slowly expanding lesions (SELs) and paramagnetic rim lesions (PRLs). OBJECTIVES: To identify the relationship between SELs and PRLs in MS, and their association with disability. METHODS: 61 people with MS (pwMS) followed retrospectively with MRI including baseline susceptibility-weighted imaging, and longitudinal T1 and T2-weighted scans. SELs were computed using deformation field maps; PRLs were visually identified. Mixed-effects models assessed differences in Expanded Disability Status Scale (EDSS) score changes between the group defined by the presence of SELs and or PRLs. RESULTS: The median follow-up time was 3.2 years. At baseline, out of 1492 lesions, 616 were classified as SELs, and 80 as PRLs. 92% of patients had ⩾ 1 SEL, 56% had ⩾ 1 PRL, while both were found in 51%. SELs compared to non-SELs were more likely to also be PRLs (7% vs. 4%, p = 0.027). PRL counts positively correlated with SEL counts (ρ= 0.28, p = 0.03). SEL + PRL + patients had greater increases in EDSS over time (beta = 0.15/year, 95% confidence interval (0.04, 0.27), p = 0.009) than SEL+PRL-patients. CONCLUSION: SELs are more numerous than PRLs in pwMS. Compared with either SELs or PRLs found in isolation, their joint occurrence was associated with greater clinical progression.

Detection of lesions in the optic nerve with magnetic resonance imaging using a 3D convolutional neural network.

BACKGROUND: Optic neuritis (ON) is one of the first manifestations of multiple sclerosis, a disabling disease with rising prevalence. Detecting optic nerve lesions could be a relevant diagnostic marker in patients with multiple sclerosis. OBJECTIVES: We aim to create an automated, interpretable method for optic nerve lesion detection from MRI scans. MATERIALS AND METHODS: We present a 3D convolutional neural network (CNN) model that learns to detect optic nerve lesions based on T2-weighted fat-saturated MRI scans. We validated our system on two different datasets (N = 107 and 62) and interpreted the behaviour of the model using saliency maps. RESULTS: The model showed good performance (68.11% balanced accuracy) that generalizes to unseen data (64.11%). The developed network focuses its attention to the areas that correspond to lesions in the optic nerve. CONCLUSIONS: The method shows robustness and, when using only a single imaging sequence, its performance is not far from diagnosis by trained radiologists with the same constraint. Given its speed and performance, the developed methodology could serve as a first step to develop methods that could be translated into a clinical setting.

T1/T2-weighted ratio in multiple sclerosis: A longitudinal study with clinical associations.

BACKGROUND: T1w/T2-w ratio has been proposed as a clinically feasible MRI biomarker to assess tissue integrity in multiple sclerosis. However, no data is available in the earliest stages of the disease and longitudinal studies analysing clinical associations are scarce. OBJECTIVE: To describe longitudinal changes in T1-w/T2-w in patients with clinically isolated syndrome (CIS) and multiple sclerosis, and to investigate their clinical associations. METHODS: T1-w/T2-w images were generated and the mean value obtained in the corresponding lesion, normal-appearing grey (NAGM) and white matter (NAWM) masks. By co-registering baseline to follow-up MRI, evolved lesions were assessed; and by placing the mask of new lesions to the baseline study, the pre-lesional tissue integrity was measured. RESULTS: We included 171 CIS patients and 22 established multiple sclerosis patients. In CIS, evolved lesions showed significant T1-w/T2-w increases compared to baseline (+7.6%, P < 0.001). T1-w/T2-w values in new lesions were lower than in pre-lesional tissue (-28.2%, P < 0.001), and pre-lesional tissue was already lower than baseline NAWM (-7.8%, P < 0.001). In CIS at baseline, higher NAGM T1-w/T2-w was associated with multiple sclerosis diagnosis, and longitudinal decreases in NAGM and NAWM T1-w/T2-w were associated with disease activity. In established multiple sclerosis, T1-w/T2-w was inversely correlated with clinical disability and disease duration. CONCLUSION: A decrease in T1-w/T2-w ratio precedes lesion formation. In CIS, higher T1-w/T2-w was associated with multiple sclerosis diagnosis. In established multiple sclerosis, lower T1-w/T2-w values were associated with clinical disability. The possible differential impact of chronic inflammation, iron deposition and demyelination should be considered to interpret these findings.

Quantitative comparison of subcortical and ventricular volumetry derived from MPRAGE and MP2RAGE images using different brain morphometry software.

OBJECTIVE: In brain volume assessment with MR imaging, it is of interest to know the effects of the pulse sequence and software used, to determine whether they provide equivalent data. The aim of this study was to compare cross-sectional volumes of subcortical and ventricular structures and their repeatability derived from MP2RAGE and MPRAGE images using MorphoBox, and FIRST or ALVIN. MATERIALS AND METHODS: MPRAGE and MP2RAGE T1-weighted images were obtained from 24 healthy volunteers. Back-to-back scans were performed in 12 of them. Volumes, coefficients of variation, concordance, and correlations were determined. RESULTS: Significant differences were found for volumes derived from MorphoBox and FIRST. Ventricular volumes determined by MorphoBox and ALVIN were similar. Differences between volumes obtained using MPRAGE and MP2RAGE were significant for a few regions. Coefficients of variation, ranged from 0.2 to 9.1%, showed a significant inverse correlation with the mean volume. There was a correlation between volume measures, but agreement was rated as poor for most regions. CONCLUSION: MP2RAGE sequences and MorphoBox are valid options for assessing subcortical and ventricular volumes, in the same way as MPRAGE and FIRST or ALVIN, accepted tools for clinical research. However, caution is needed when comparing volumes obtained with different tools.

Assessment of brain volumes obtained from MP-RAGE and MP2RAGE images, quantified using different segmentation methods.

OBJECTIVE: For clinical purposes and research projects in neurological disease, it is of interest to evaluate the performance and comparability of available sequences and software packages for brain volume assessment to determine whether they provide equivalent results. This study compares cross-sectional brain volume values derived from images obtained with MP-RAGE or MP2RAGE sequences, using SIENA/X, SPM, or MorphoBox. MATERIALS AND METHODS: MP-RAGE and MP2RAGE T1-weighted images were obtained from 24 healthy volunteers. Back-to-back scans were performed in 12 of them. Brain volumes, coefficients of variation, and concordance coefficients were determined. RESULTS: Significant differences were found for most brain volumes derived from MP-RAGE and MP2RAGE images. MP2RAGE-derived measures showed a non-significant trend to larger coefficients of variation. There were statistical differences between brain volumes determined with the three software packages, whereas coefficients of variation were comparable for most brain volumes. Correlation and concordance values were lower for CSF and brain parenchyma fraction measures. CONCLUSION: The results obtained advise caution when comparing brain volumes obtained by different sequences and software packages. Of note, for most brain volume measures, the MP2RAGE and MorphoBox coefficients of variation were similar to those obtained with MP-RAGE, SIENA/X or SPM, accepted tools for clinical research.

A validation study of manual atrophy measures in patients with Multiple Sclerosis.

PURPOSE: Manual measures such as corpus callosum index, normalized corpus callosum area, and width of the third ventricle are potential biomarkers for brain atrophy. In this work, we investigate their suitability to assess the neurodegenerative component of multiple sclerosis (MS) by comparing them to volumetric measures and expanded disability status scale (EDSS). METHODS: Fifty-eight patients with a clinically isolated syndrome, 48 MS patients treated with interferon ß, and 26 treated with natalizumab underwent a brain MRI at baseline and after 1 year. Manual measures were evaluated by two observers using Jim v.6.0 at both time points. Volumetric tools (SIENA/x and Freesurfer) were used to calculate normalized brain volume, brain parenchymal fraction, annualized percentage of brain volume change, corpus callosum volume, ventricle volume, and volume of the third ventricle. Statistical analyses were performed with SPSS v.13. RESULTS: Usage of corpus callosum volume and third ventricle volume to validate normalized corpus callosum area and width of the third ventricle, respectively, showed very good correlations (r = 0.85, r = 0.83; p < 0.01). Width of the third ventricle, corpus callosum index, and normalized corpus callosum area correlations were significant with EDSS in all patients and moderate to strong with normalized brain volume and brain parenchymal fraction in natalizumab-treated patients (respectively r = - 0.54, r = - 0.61; r = 0.55, r = 0.67; and r = 0.58, r = 0.67; with p < 0.05). CONCLUSION: Width of the third ventricle and normalized corpus callosum area seem the more robust manual measures regarding correlation with volumetric measures and EDSS, especially in patients with more advanced disease.

Brain regional volume estimations with NeuroQuant and FIRST: a study in patients with a clinically isolated syndrome.

PURPOSE: Brain volume estimates from magnetic resonance images (MRIs) are of great interest in multiple sclerosis, and several automated tools have been developed for this purpose. The goal of this study was to assess the agreement between two tools, NeuroQuant® (NQ) and FMRIB's Integrated Registration Segmentation Tool (FIRST), for estimating overall and regional brain volume in a cohort of patients with a clinically isolated syndrome (CIS). In addition, white matter lesion volume was estimated with NQ and the Lesion Segmentation Toolbox (LST). METHODS: One hundred fifteen CIS patients were analysed. Structural images were acquired on a 3.0-T system. The volume agreement between methods (by estimation of the intraclass correlation coefficient) was calculated for the right and left thalamus, caudate, putamen, pallidum, hippocampus, and amygdala, as well as for the total intracranial volume and white matter lesion volume. RESULTS: In general, the estimated volumes were larger by NQ than FIRST, except for the pallidum. Agreement was low (ICC < 0.40) for the smaller structures (amygdala and pallidum) and fair to good (ICC > 0.40) for the remaining ones. Agreement was fair for lesion volume (ICC = 0.61), with NQ estimates lower than LST. CONCLUSIONS: Agreement between NQ and FIRST brain volume estimates depends on the size of the structure of interest, with larger volumes achieving better agreement. In addition, concordance between the two tools does seem to be dependent on the presence of brain lesions.