Multiple Sclerosis Data Set Research Articles

Analisis Klasifikasi Penyakit Multiple Sclerosis Menggunakan Algoritma Logistic Regression dan SVM

Health is the most important aspect to support daily activities. Of course, by having a healthy body, everyone can carry out various activities comfortably and calmly. Every individual certainly has a strong instinct to live a healthy life and be free from disease, one of which is by increasing the body's immunity. Multiple sclerosis (multiple sclerosis/MS) is a neurodegenerative autoimmune disease that affects the central nervous system. The affliction of MS is characterized by chronic inflammation, demyelination, gliosis, and neuronal death. The symptoms faced by MS patients are unpredictable, so there is a need for a classification related to the disease. Therefore, a classification study was carried out using the logistic regression algorithm and SVM. The method used in this research is a literature study with the Python programming language. The results of this study indicate that the SVM algorithm has a high accuracy rate of 88.33% of the logistic regression algorithm. So it can be concluded from this study that the SVM method has good performance for processing multiple sclerosis datasets.

Exploring Shared Genetic Signatures of Alzheimer’s Disease and Multiple Sclerosis: A Bioinformatic Analysis Study

Introduction: Many clinical studies reported the coexistence of Alzheimer’s disease (AD) and multiple sclerosis (MS), but the common molecular signature between AD and MS remains elusive. The purpose of our study was to explore the genetic linkage between AD and MS through bioinformatic analysis, providing new insights into the shared signatures and possible pathogenesis of two diseases. Methods: The common differentially expressed genes (DEGs) were determined between AD and MS from datasets obtained from Gene Expression Omnibus (GEO) database. Further, functional and pathway enrichment analysis, protein-protein interaction network construction, and identification of hub genes were carried out. The expression level of hub genes was validated in two other external AD and MS datasets. Transcription factor (TF)-gene interactions and gene-miRNA interactions were performed in NetworkAnalyst. Finally, receiver operating characteristic (ROC) curve analysis was applied to evaluate the predictive value of hub genes. Results: A total of 75 common DEGs were identified between AD and MS. Functional and pathway enrichment analysis emphasized the importance of exocytosis and synaptic vesicle cycle, respectively. Six significant hub genes, including CCL2, CD44, GFAP, NEFM, STXBP1, and TCEAL6, were identified and verified as common hub genes shared by AD and MS. FOXC1 and hsa-mir-16-5p are the most common TF and miRNA in regulating hub genes, respectively. In the ROC curve analysis, all hub genes showed good efficiency in helping distinguish patients from controls. Conclusion: Our study first identified a common genetic signature between AD and MS, paving the road for investigating shared mechanism of AD and MS.

Repeated forms, testing intervals, and SDMT performance in a large multiple sclerosis dataset

BackgroundThe Symbol Digit Modalities Test (SDMT), the most reliable and sensitive measure of cognition in people with multiple sclerosis (PwMS), is increasingly used in clinical trials and care. ObjectivesWe aimed to establish how SDMT performance is influenced by repeating forms and frequency of use in PwMS. MethodsA retrospective analysis was completed on a large database of PwMS (n = 740) with multiple SDMT administrations. Change in SDMT performance was analyzed, accounting for frequency of tests and utilization of alternate- versus same-form conditions. ResultsSDMT administrations ranged from 2 to 14 per subject over a mean (SD) of 5.9 (4.5) years. Accounting for demographics, the mixed effects model revealed a significant main effect of SDMT exposures (1.8 point improvement per repetition, p = 0.001) and an interaction between time since previous SDMT and whether the same test form was administered in the previous administration (estimate=-1.1, p = 0.037). As well, SDMT decline is observed when testing intervals exceed two years (F = 9.69, p<0.001). ConclusionImprovements in SDMT performance with repeated exposure, likely reflecting practice effects, were greatest when repeating the same SDMT form over briefer intervals. We recommend the use of alternate forms or analogous versions of timed symbol-digit coding particularly where samples are saturated with many administrations.

Multiple Sclerosis Biomarker Candidates Revealed by Cell-Type-Specific Interactome Analysis.

Multiple sclerosis (MS) is a demyelinating disorder that affects multiple regions of the central nervous system such as the brain, spinal cord, and optic nerves. Susceptibility to MS, as well as disease progression rates, displays marked patient-to-patient variability. To date, biomarkers that forecast differences in clinical phenotypes and outcomes have been limited. In this context, cell-type-specific interactome analyses offer important prospects and hope for novel diagnostics and therapeutics. We report here an original study using bioinformatic analysis of MS data sets that revealed interaction profiles as well as specific hub proteins in white matter (WM) and gray matter (GM) that appear critical for disease mechanisms. First, cell-type-specific interactome analyses suggested that while interactions within the WM were focused on oligodendrocytes, interactions within the GM were mostly neuron centric. Second, hub proteins such as APP, EGLN3, PTEN, and LRRK2 were identified to be differentially regulated in MS data sets. Lastly, a comparison of the brain and peripheral blood samples identified biomarker candidates such as NRGN, CRTC1, CDC42, and IFITM3 to be differentially expressed in different types of MS. These findings offer a unique cell-type-specific cell-to-cell interaction network in MS and identify potential biomarkers by comparative analysis of the brain and the blood transcriptomics. From a study design and methodology perspective, we suggest that the cell-type-specific interactome analysis is an important systems science frontier that might offer new insights on other neurodegenerative and brain disorders as well.

Hidden Markov Model and multifractal method-based predictive quantization complexity models vis-á-vis the differential prognosis and differentiation of Multiple Sclerosis’ subgroups

Hidden Markov Model (HMM) is a stochastic process where implicit or latent stochastic processes can be inferred indirectly through a sequence of observed states. HMM as a mathematical model for uncertain phenomena is applicable for the description and computation of complex dynamical behaviours enabling the mathematical formulation of neural dynamics across spatial and temporal scales. The human brain with its fractal structure demonstrates complex dynamics and fractals in the brain are characterized by irregularity, singularity and self-similarity in terms of form at different observation levels, making detection difficult as observations in real-time occurrences can be time variant, discrete, continuous or noisy. Multiple Sclerosis (MS) is an autoimmune degenerative disease with time and space related dissemination, leading to neuronal apoptosis, coupled with some subtle features that could be overlooked by physicians. This study, through the proposed integrated approach with multi-source complex spatial data, aims to attain accurate prediction, diagnosis and prognosis of MS subgroups by HMM with Viterbi algorithm and Forward–Backward algorithm as the dynamic and efficient products of knowledge-based and Artificial Intelligence (AI)-based systems within the framework of precision medicine. Multifractal Bayesian method (MFM) accordingly applied to identify and eliminate “insignificant” irregularities while maintaining “significant” singularities. An efficient modelling of HMM is proposed to diagnose and predict the course of MS while using MFM method. Unlike the methods employed in previous studies, our proposed integrated novel method encompasses the subsequent approaches based on reliable MS dataset (X) collected: (i) MFM method was applied (X) to MS dataset to characterize the irregular, self-similar and significant attributes, thus, attributes with “insignificant” irregularities were eliminated and “significant” singularities were maintained. MFM-MS dataset (Xˆ) was generated. (ii) The continuous values in the MS dataset (X) and MFM-MS dataset (Xˆ) were converted into discrete values through vector quantization method of the HMM (iii) Through transitional matrices, different observation matrices were computed from the both datasets. (v) Computational complexity has been computed for both datasets. (vi) The results of the HMM models based on observation matrices obtained from both datasets were compared. In terms of the integrated HMM model proposed and the MS dataset handled, no earlier work exists in the literature. The experimental results demonstrate the applicability and accuracy of our novel proposed integrated method, HMM and Multifractal (HMM-MFM) method, for the application to the MS dataset (X). Compared with conventional methods, our novel method has achieved more superiority regarding extracting subtle and hidden details, which are significant for distinguishing different dynamic and complex systems including engineering and other related applied sciences. Thus, we have aimed at pointing a new frontier by providing a novel alternative mathematical model to facilitate the critical decision-making, management and prediction processes among the related areas in chaotic, dynamic complex systems with intricate and transient states.

Bi-directional Synthesis of Pre- and Post-contrast MRI via Guided Feature Disentanglement.

Magnetic resonance imaging (MRI) with gadolinium contrast is widely used for tissue enhancement and better identification of active lesions and tumors. Recent studies have shown that gadolinium deposition can accumulate in tissues including the brain, which raises safety concerns. Prior works have tried to synthesize post-contrast T1-weighted MRIs from pre-contrast MRIs to avoid the use of gadolinium. However, contrast and image representations are often entangled during the synthesis process, resulting in synthetic post-contrast MRIs with undesirable contrast enhancements. Moreover, the synthesis of pre-contrast MRIs from post-contrast MRIs which can be useful for volumetric analysis is rarely investigated in the literature. To tackle pre- and post- contrast MRI synthesis, we propose a BI-directional Contrast Enhancement Prediction and Synthesis (BICEPS) network that enables disentanglement of contrast and image representations via a bi-directional image-to-image translation(I2I)model. Our proposed model can perform both pre-to-post and post-to-pre contrast synthesis, and provides an interpretable synthesis process by predicting contrast enhancement maps from the learned contrast embedding. Extensive experiments on a multiple sclerosis dataset demonstrate the feasibility of applying our bidirectional synthesis and show that BICEPS outperforms current methods.

Association of Age With Contrast-Enhancing Lesions Across the Multiple Sclerosis Disease Spectrum.

To investigate the association of age and the presence of contrast-enhancing lesions (CELs) on cranial MRI scans in different disease courses of multiple sclerosis (MS), we describe the frequency of CELs as a function of age in 4 large randomized controlled trial (RCT) datasets. Using original trial data from CombiRx (Combination Therapy in Patients With Relapsing-Remitting Multiple Sclerosis; clinicaltrials.gov identifier NCT00211887), a trial in relapsing-remitting MS; ASCEND (A Clinical Study of the Efficacy of Natalizumab on Reducing Disability Progression in Participants With Secondary Progressive Multiple Sclerosis; clinicaltrials.gov identifier NCT01416181), a trial in secondary progressive MS; and the 2 primary progressive MS trials PROMISE and INFORMS; clinicaltrials.gov identifier NCT00731692), we describe the occurrence of CELs per age group at baseline for the entire trial cohort and at 1 year follow-up in the treatment arms. CombiRx included 1,008, ASCEND 889, PROMISE 943, and INFORMS 970 participants. At baseline, CEL frequency differed between datasets according to disease course: 39.6% of CombiRx, 23.9% of ASCEND, 14.0% of PROMISE, and 12.3% of INFORMS participants had CELs. This distribution by disease course was largely preserved within each age group. In all datasets, there was an almost linear decrease of the percentage of participants with CELs with advancing age. After 1 year of experimental treatment, CEL occurrence was reduced in all trial datasets, and almost absent in ASCEND. The decrease of CEL occurrence with advancing age was preserved in CombiRx, PROMISE, and INFORMS after 1 year of treatment. We investigated the association of the baseline factors age, disease duration, sex, and EDSS with having CELs at baseline with multivariable binary logistic regression models. Age was the only characteristic associated with the risk of CELs at baseline in all datasets, with higher age associated with a lower risk of CELs (odds ratios for having CELs at baseline per year increase in age: CombiRx: 0.96, 95% confidence interval [CI] 0.95-0.98; ASCEND: 0.94, 95% CI 0.92-0.97; PROMISE: 0.94, 95% CI 0.91-0.96; INFORMS: 0.97, 95% CI 0.94-0.99). Our analysis of 4 large, well-characterized RCT datasets shows that the association of age and CEL occurrence is a general phenomenon across the spectrum of MS disease courses. Our findings should be replicated in real-world MS datasets.

Characterisation of MS phenotypes across the age span using a novel data set integrating 34 clinical trials (NO.MS cohort): Age is a key contributor to presentation

Background: The Oxford Big Data Institute, multiple sclerosis (MS) physicians and Novartis aim to address unresolved questions in MS with a novel comprehensive clinical trial data set. Objective: The objective of this study is to describe the Novartis–Oxford MS (NO.MS) data set and to explore the relationships between age, disease activity and disease worsening across MS phenotypes. Methods: We report key characteristics of NO.MS. We modelled MS lesion formation, relapse frequency, brain volume change and disability worsening cross-sectionally, as a function of patients’ baseline age, using phase III study data (≈8000 patients). Results: NO.MS contains data of ≈35,000 patients (>200,000 brain images from ≈10,000 patients), with >10 years follow-up. (1) Focal disease activity is highest in paediatric patients and decreases with age, (2) brain volume loss is similar across age and phenotypes and (3) the youngest patients have the lowest likelihood (<25%) of disability worsening over 2 years while risk is higher (25%–75%) in older, disabled or progressive MS patients. Young patients benefit most from treatment. Conclusion: NO.MS will illuminate questions related to MS characterisation, progression and prognosis. Age modulates relapse frequency and, thus, the phenotypic presentation of MS. Disease worsening across all phenotypes is mediated by age and appears to some extent be independent from new focal inflammatory activity.

A Framework for Efficient N-Way Interaction Testing in Case/Control Studies With Categorical Data.

Goal: Most common diseases are influenced by multiple gene interactions and interactions with the environment. Performing an exhaustive search to identify such interactions is computationally expensive and needs to address the multiple testing problem. A four-step framework is proposed for the efficient identification of n-Way interactions. Methods: The framework was applied on a Multiple Sclerosis dataset with 725 subjects and 147 tagging SNPs. The first two steps of the framework are quality control and feature selection. The next step uses clustering and binary encodes the features. The final step performs the n-Way interaction testing. Results: The feature space was reduced to 7 SNPs and using the proposed binary encoding, more 2-SNP and 3-SNP interactions were identified compared to using the initial encoding. Conclusions: The framework selects informative features and with the proposed binary encoding it is able to identify more n-way interactions by increasing the power of the statistical analysis.

ALL-Net: Anatomical information lesion-wise loss function integrated into neural network for multiple sclerosis lesion segmentation

Accurate detection and segmentation of multiple sclerosis (MS) brain lesions on magnetic resonance images are important for disease diagnosis and treatment. This is a challenging task as lesions vary greatly in size, shape, location, and image contrast. The objective of our study was to develop an algorithm based on deep convolutional neural network integrated with anatomic information and lesion-wise loss function (ALL-Net) for fast and accurate automated segmentation of MS lesions. Distance transformation mapping was used to construct a convolutional module that encoded lesion-specific anatomical information. To overcome the lesion size imbalance during network training and improve the detection of small lesions, a lesion-wise loss function was developed in which individual lesions were modeled as spheres of equal size. On the ISBI-2015 longitudinal MS lesion segmentation challenge dataset (19 subjects in total), ALL-Net achieved an overall score of 93.32 and was amongst the top performing methods. On the larger Cornell MS dataset (176 subjects in total), ALL-Net significantly improved both voxel-wise metrics (Dice improvement of 3.9% to 35.3% with p-values ranging from p < 0.01 to p < 0.0001, and AUC of voxel-wise precision-recall curve improvement of 2.1% to 29.8%) and lesion-wise metrics (lesion-wise F1 score improvement of 12.6% to 29.8% with all p-values p < 0.0001, and AUC of lesion-wise ROC curve improvement of 1.4% to 20.0%) compared to leading publicly available MS lesion segmentation tools.

Validating atlas-based lesion disconnectomics in multiple sclerosis: A retrospective multi-centric study

The translational potential of MR-based connectivity modelling is limited by the need for advanced diffusion imaging, which is not part of clinical protocols for many diseases. In addition, where diffusion data is available, brain connectivity analyses rely on tractography algorithms which imply two major limitations. First, tracking algorithms are known to be sensitive to the presence of white matter lesions and therefore leading to interpretation pitfalls and poor inter-subject comparability in clinical applications such as multiple sclerosis. Second, tractography quality is highly dependent on the acquisition parameters of diffusion sequences, leading to a trade-off between acquisition time and tractography precision.Here, we propose an atlas-based approach to study the interplay between structural disconnectivity and lesions without requiring individual diffusion imaging. In a multi-centric setting involving three distinct multiple sclerosis datasets (containing both 1.5 T and 3 T data), we compare our atlas-based structural disconnectome computation pipeline to disconnectomes extracted from individual tractography and explore its clinical utility for reducing the gap between radiological findings and clinical symptoms in multiple sclerosis. Results using topological graph properties showed that overall, our atlas-based disconnectomes were suitable approximations of individual disconnectomes from diffusion imaging. Small-worldness was found to decrease for larger total lesion volumes thereby suggesting a loss of efficiency in brain connectivity of MS patients. Finally, the global efficiency of the created brain graph, combined with total lesion volume, allowed to stratify patients into subgroups with different clinical scores in all three cohorts.

COVID-19 in ocrelizumab-treated people with multiple sclerosis

BackgroundThere are limited data on the impact of coronavirus disease 2019 (COVID-19) caused by the severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) on people with multiple sclerosis (MS). ObjectiveTo better understand SARS-CoV-2 infection in ocrelizumab-treated people with MS. MethodsInternal Roche/Genentech data sources: Cases of COVID-19 from ongoing Roche/Genentech clinical trials and from post-marketing use of ocrelizumab until July 31, 2020 were identified and assessed using descriptive statistics.External real-world data (RWD) source: An MS COVID-19 cohort and an ocrelizumab-treated MS COVID-19 cohort were identified and assessed from the OPTUMⓇ de-identified COVID-19 electronic health record (EHR) database. ResultsRoche/Genentech clinical trial data: There were 51 (1.3%) suspected or confirmed cases of COVID-19 identified from 4,000 patients ongoing in 10 Roche/Genentech clinical trials. Of these, 26 (51%) were confirmed COVID-19 and 25 (49%) were suspected COVID-19. Sixteen (31.4%) patients were hospitalized. COVID-19 severity was mild to moderate in most patients (35, 68.6%). Ten (19.6%) patients had severe disease and there were three (5.9%) fatal cases. Most patients (43, 84.3%) recovered or were recovering. There was no association apparent between duration of exposure to ocrelizumab and COVID-19. Among COVID-19 patients with previous serum immunoglobulin status (27/51, 52.9%), all (27/27, 100%) had IgG levels within the normal range.Roche/Genentech post-marketing safety database data: There were 307 post-marketing cases of COVID-19 in the Roche/Genentech global safety database. Of these, 263 (85.7%) were confirmed and 44 (14.3%) were suspected COVID-19. 100 (32.6%) patients were hospitalized. COVID-19 was asymptomatic, mild or moderate in 143 (46.6%) patients, severe in 52 (16.9%) patients, and critical in 15 (4.9%) patients. There were 17 (5.5%) fatal cases. Information on severity was not reported in 80 (26.1%) cases. Most patients (211, 68.7%) recovered or were recovering at the time of the report.External RWD data source: As of July 13, 2020, the OPTUMⓇ database included EHRs for almost 1.2 million patients with suspected COVID-19, 130,500 of whom met the criteria for confirmed/clinically diagnosed COVID-19. A total of 357 patients with MS with confirmed COVID-19 were identified. Forty-eight (13.4%) were treated with ocrelizumab, of whom 12 (25.0%) were hospitalized and one died (2.1%). Similar rates of hospitalization, invasive ventilation, and death were observed in the ocrelizumab-treated and non-ocrelizumab-treated MS cohorts.Across the Roche/Genentech and RWD sources assessed, age, male sex, and the presence of comorbidities such as hypertension were associated with a more severe disease course of COVID-19. There was a higher number of comorbidities present in hospitalized versus non-hospitalized patients. ConclusionsThis assessment provides evidence that COVID-19 in ocrelizumab-treated people with MS is predominantly mild to moderate in severity with most patients not requiring hospitalization; in line with data reported from the general population and MS datasets. Risk factors known to be associated with severe COVID-19 outcomes in the general population also appear to influence COVID-19 severity in ocrelizumab-treated people with MS. Case fatality rates for ocrelizumab-treated people with MS were within published ranges for the general population and other MS cohorts.

Treatment Response Score to Glatiramer Acetate or Interferon Beta-1a.

To compare the effectiveness of glatiramer acetate (GA) vs intramuscular interferon beta-1a (IFN-β-1a), we applied a previously published statistical method aimed at identifying patients' profiles associated with efficacy of treatments. Data from 2 independent multiple sclerosis datasets, a randomized study (the Combination Therapy in Patients With Relapsing-Remitting Multiple Sclerosis [CombiRx] trial, evaluating GA vs IFN-β-1a) and an observational cohort extracted from MSBase, were used to build and validate a treatment response score, regressing annualized relapse rates (ARRs) on a set of baseline predictors. The overall ARR ratio of GA to IFN-β-1a in the CombiRx trial was 0.72. The response score (made up of a linear combination of age, sex, relapses in the previous year, disease duration, and Expanded Disability Status Scale score) detected differential response of GA vs IFN-β-1a: in the trial, patients with the largest benefit from GA vs IFN-β-1a (lower score quartile) had an ARR ratio of 0.40 (95% confidence interval [CI] 0.25-0.63), those in the 2 middle quartiles of 0.90 (95% CI 0.61-1.34), and those in the upper quartile of 1.14 (95% CI 0.59-2.18) (heterogeneity p = 0.012); this result was validated on MSBase, with the corresponding ARR ratios of 0.58 (95% CI 0.46-0.72), 0.92 (95% CI 0.77-1.09,) and 1.29 (95% CI 0.97-1.71); heterogeneity p < 0.0001). We demonstrate the possibility of a criterion, based on patients' characteristics, to choose whether to treat with GA or IFN-β-1a. This result, replicated on an independent real-life cohort, may have implications for clinical decisions in everyday clinical practice.

Exploring uncertainty measures in deep networks for Multiple sclerosis lesion detection and segmentation.

Deep learning networks have recently been shown to outperform other segmentation methods on various public, medical-image challenge datasets, particularly on metrics focused on large pathologies. For diseases such as Multiple Sclerosis (MS), however, monitoring all the focal lesions visible on MRI sequences, even very small ones, is essential for disease staging, prognosis, and evaluating treatment efficacy. Small lesion segmentation presents significant challenges to popular deep learning models. This, coupled with their deterministic predictions, hinders their clinical adoption. Uncertainty estimates for these predictions would permit subsequent revision by clinicians. We present the first exploration of multiple uncertainty estimates based on Monte Carlo (MC) dropout(Gal and Ghahramani, 2016) in the context of deep networks for lesion detection and segmentation in medical images. Specifically, we develop a 3D MS lesion segmentation CNN, augmented to provide four different voxel-based uncertainty measures based on MC dropout. We train the network on a proprietary, large-scale, multi-site, multi-scanner, clinical MS dataset, and compute lesion-wise uncertainties by accumulating evidence from voxel-wise uncertainties within detected lesions. We analyze the performance of voxel-based segmentation and lesion-level detection by choosing operating points based on the uncertainty. Uncertainty filtering improves both voxel and lesion-wise TPR and FDR on remaining, certain predictions compared to sigmoid-based TPR/FDR curves. Small lesions and lesion-boundaries are the most uncertain regions, which is consistent with human-rater variability.

An image classification framework exploring the capabilities of extreme learning machines and artificial bee colony

A hybridized image classification strategy is proposed based on discrete wavelet transform, artificial bee colony (ABC) and extreme learning machine (ELM). The proposed methodology works in three phases: (a) in preprocessing phase, images are decomposed and features are extracted from images using bi-orthogonal wavelet functions; (b) secondly, modified ABC (MABC) optimization algorithm is proposed to determine the optimal parameters such as hidden layer weights and biases to be used by ELM for classification; (c) the ELM in the third phase has been trained and tested with three brain image datasets for different diseases along with normal brain images. The performance recognition of the proposed MABC-ELM in terms of accuracy, rate of per-image classification and speedup has been made with variants of ELM such as ELM, ABC-ELM and MABC-ELM and also with MLPNN, naive Bayesian, linear regression classifiers. Finally, the percentage of accuracy observed by the proposed MABC-ELM, for acute stroke-speech arrest, glioma and multiple sclerosis datasets, is 90%, 90% and 100% with eight hidden nodes in the ELM architecture, and it can be concluded that MABC-ELM gives better generalization performance, more compact network architecture and the hybridization of ELM with modified ABC is worth investigated.

Accurate, rapid and reliable, fully automated MRI brainstem segmentation for application in multiple sclerosis and neurodegenerative diseases.

Neurodegenerative disorders, such as Alzheimer's disease (AD) and progressive forms of multiple sclerosis (MS), can affect the brainstem and are associated with atrophy that can be visualized by MRI. Anatomically accurate, large-scale assessments of brainstem atrophy are challenging due to lack of automated, accurate segmentation methods. We present a novel method for brainstem volumetry using a fully-automated segmentation approach based on multi-dimensional gated recurrent units (MD-GRU), a deep learning based semantic segmentation approach employing a convolutional adaptation of gated recurrent units. The neural network was trained on 67 3D-high resolution T1-weighted MRI scans from MS patients and healthy controls (HC) and refined using segmentations of 20 independent MS patients' scans. Reproducibility was assessed in MR test-retest experiments in 33 HC. Accuracy and robustness were examined by Dice scores comparing MD-GRU to FreeSurfer and manual brainstem segmentations in independent MS and AD datasets. The mean %-change/SD between test-retest brainstem volumes were 0.45%/0.005 (MD-GRU), 0.95%/0.009 (FreeSurfer), 0.86%/0.007 (manually edited segmentations). Comparing MD-GRU to manually edited segmentations the mean Dice scores/SD were: 0.97/0.005 (brainstem), 0.95/0.013 (mesencephalon), 0.98/0.006 (pons), 0.95/0.015 (medulla oblongata). Compared to the manual gold standard, MD-GRU brainstem segmentations were more accurate than FreeSurfer segmentations (p < .001). In the multi-centric acquired AD data, the mean Dice score/SD for the MD-GRU-manual segmentation comparison was 0.97/0.006. The fully automated brainstem segmentation method MD-GRU provides accurate, highly reproducible, and robust segmentations in HC and patients with MS and AD in 200 s/scan on an Nvidia GeForce GTX 1080 GPU and shows potential for application in large and longitudinal datasets.

Dynamic 11C-PiB PET Shows Cerebrospinal Fluid Flow Alterations in Alzheimer Disease and Multiple Sclerosis.

Cerebrospinal fluid (CSF) plays an important role in solute clearance and maintenance of brain homeostasis. 11C-Pittsburgh compound B (PiB) PET was recently proposed as a tool for detection of CSF clearance alterations in Alzheimer disease. The current study investigates the magnitude of 11C-PiB PET signal in the lateral ventricles of an independent group of Alzheimer and mild cognitive impairment subjects. We have also evaluated multiple sclerosis as a model of disease with CSF clearance alterations without amyloid-β tissue accumulation. Methods: A set of 11 Alzheimer and 12 mild cognitive impairment subjects and a set of 20 multiple sclerosis subjects with matched controls underwent MRI and dynamic 11C-PiB PET. Lateral ventricle regions of interest were generated manually from MRI data. PET data were analyzed using cerebellum or a supervised reference region for the Alzheimer and multiple sclerosis data sets, respectively. The magnitude of 11C-PiB signal in the lateral ventricles was calculated as area under the curve from 35 to 80 min and SUV ratio (SUVR) from 50 to 70 min. Compartmental modeling analysis was performed on a separate data set containing 11 Alzheimer and matched control subjects; this analysis included an arterial input function, to further understand the kinetics of the lateral ventricular 11C-PiB signal. Results: ANOVA revealed significant group differences in lateral ventricular SUVR across the Alzheimer, mild cognitive impairment, and healthy control groups (P = 0.004). Pairwise comparisons revealed significantly lower lateral ventricular SUVR in Alzheimer subjects than in healthy controls (P < 0.001) or mild cognitive impairment subjects (P = 0.029). Lateral ventricular SUVR was significantly lower in multiple sclerosis subjects than in healthy controls (P = 0.008). Compartmental modeling analysis revealed significantly lower uptake rates of 11C-PiB signal from blood (P = 0.005) and brain tissue (P = 0.004) to the lateral ventricles and significantly lower 11C-PiB signal clearance out of the lateral ventricles (P = 0.002) in Alzheimer subjects than in healthy controls. Conclusion: These results indicate that dynamic 11C-PiB PET can be used to observe pathologic changes in CSF dynamics. We have replicated previous work demonstrating CSF clearance deficits in Alzheimer disease associated with amyloid-β deposits and have extended the observations to include ventricular CSF clearance deficits in mild cognitive impairment and multiple sclerosis.

One-shot domain adaptation in multiple sclerosis lesion segmentation using convolutional neural networks.

In recent years, several convolutional neural network (CNN) methods have been proposed for the automated white matter lesion segmentation of multiple sclerosis (MS) patient images, due to their superior performance compared with those of other state-of-the-art methods. However, the accuracies of CNN methods tend to decrease significantly when evaluated on different image domains compared with those used for training, which demonstrates the lack of adaptability of CNNs to unseen imaging data. In this study, we analyzed the effect of intensity domain adaptation on our recently proposed CNN-based MS lesion segmentation method. Given a source model trained on two public MS datasets, we investigated the transferability of the CNN model when applied to other MRI scanners and protocols, evaluating the minimum number of annotated images needed from the new domain and the minimum number of layers needed to re-train to obtain comparable accuracy. Our analysis comprised MS patient data from both a clinical center and the public ISBI2015 challenge database, which permitted us to compare the domain adaptation capability of our model to that of other state-of-the-art methods. In both datasets, our results showed the effectiveness of the proposed model in adapting previously acquired knowledge to new image domains, even when a reduced number of training samples was available in the target dataset. For the ISBI2015 challenge, our one-shot domain adaptation model trained using only a single case showed a performance similar to that of other CNN methods that were fully trained using the entire available training set, yielding a comparable human expert rater performance. We believe that our experiments will encourage the MS community to incorporate its use in different clinical settings with reduced amounts of annotated data. This approach could be meaningful not only in terms of the accuracy in delineating MS lesions but also in the related reductions in time and economic costs derived from manual lesion labeling.

Association of Inflammation and Disability Accrual in Patients With Progressive-Onset Multiple Sclerosis

The role of inflammatory disease activity as a determinant of disability in progressive-onset multiple sclerosis (MS) remains contested. To examine the association of superimposed relapses in progressive-onset MS on disease outcomes. Observational cohort study from MSBase, a prospectively collected, international database. Data were collected between January 1995 and February 2017. Analyses began in February 2017. From 44 449 patients at time of extraction, 1419 eligible patients (31.9%) were identified for analysis. Inclusion criteria consisted of primary progressive MS (PPMS) or progressive-relapsing MS (PRMS), adult-onset disease, and minimum data set (including ≥3 visits with disability recorded, ≥3 months between second and last visit). Data were analyzed using multivariable regression models (Andersen-Gill) with mixed effects. Two sensitivity analyses to exclude both relapse-related disability progression and bout-onset progressive MS were performed. Grouped according to presence or absence of relapse, defined as an acute episode of clinical worsening. Quantifiable disability change or correlation on imaging was not required to confirm relapse. Cumulative hazard of disability progression. Patients with PRMS were younger than those with PPMS (mean [SD] age, 46 [15] vs 51 [10] years, Cohen d = 0.40) and demonstrated a mean lower Expanded Disability Status Scale score (mean [SD] score, 4.0 [3] vs 4.5 [2.5], Cohen d = 0.28) at inclusion. The ratio of men to women was similar in the PRMS and PPMS groups (252:301 vs 394:472). The overall mean (SD) age was 48 (11) years for men and 50 (10) years for women. Likelihood of confirmed disability progression was lower in patients with superimposed relapses (hazard ratio [HR], 0.83; 95% CI, 0.74-0.94; P = .003). Proportion of follow-up time spent on disease-modifying therapy significantly reduced the hazard of confirmed disability progression in the cohort with relapse (HR, 0.96; 95% CI, 0.94-0.99; P = .01) but not in those without relapse (HR, 1.02; 95% CI, 0.99-1.05; P = .26). When accounting for relapse-related progression, the association of disease-modifying therapy in the cohort with superimposed relapse was no longer observed (HR, 1.10; 95% CI, 0.96-1.24; P = .16). In progressive-onset MS, superimposed relapses are associated with a lower risk of confirmed disability progression. This is most likely attributed to the association of disease-modifying therapy with the prevention of relapse-related disability accrual in patients with superimposed relapse. These findings suggest that inflammatory relapses are an important and modifiable determinant of disability accrual in progressive-onset disease.

Performance of five research-domain automated WM lesion segmentation methods in a multi-center MS study

Background and PurposeIn vivoidentification of white matter lesions plays a key-role in evaluation of patients with multiple sclerosis (MS). Automated lesion segmentation methods have been developed to substitute manual outlining, but evidence of their performance in multi-center investigations is lacking. In this work, five research-domain automated segmentation methods were evaluated using a multi-center MS dataset. Methods70 MS patients (median EDSS of 2.0 [range 0.0–6.5]) were included from a six-center dataset of the MAGNIMS Study Group (www.magnims.eu) which included 2D FLAIR and 3D T1 images with manual lesion segmentation as a reference. Automated lesion segmentations were produced using five algorithms: Cascade; Lesion Segmentation Toolbox (LST) with both the Lesion growth algorithm (LGA) and the Lesion prediction algorithm (LPA); Lesion-Topology preserving Anatomical Segmentation (Lesion-TOADS); and k-Nearest Neighbor with Tissue Type Priors (kNN-TTP). Main software parameters were optimized using a training set (N = 18), and formal testing was performed on the remaining patients (N = 52). To evaluate volumetric agreement with the reference segmentations, intraclass correlation coefficient (ICC) as well as mean difference in lesion volumes between the automated and reference segmentations were calculated. The Similarity Index (SI), False Positive (FP) volumes and False Negative (FN) volumes were used to examine spatial agreement. All analyses were repeated using a leave-one-center-out design to exclude the center of interest from the training phase to evaluate the performance of the method on ‘unseen’ center. ResultsCompared to the reference mean lesion volume (4.85 ± 7.29 mL), the methods displayed a mean difference of 1.60 ± 4.83 (Cascade), 2.31 ± 7.66 (LGA), 0.44 ± 4.68 (LPA), 1.76 ± 4.17 (Lesion-TOADS) and −1.39 ± 4.10 mL (kNN-TTP). The ICCs were 0.755, 0.713, 0.851, 0.806 and 0.723, respectively. Spatial agreement with reference segmentations was higher for LPA (SI = 0.37 ± 0.23), Lesion-TOADS (SI = 0.35 ± 0.18) and kNN-TTP (SI = 0.44 ± 0.14) than for Cascade (SI = 0.26 ± 0.17) or LGA (SI = 0.31 ± 0.23). All methods showed highly similar results when used on data from a center not used in software parameter optimization. ConclusionThe performance of the methods in this multi-center MS dataset was moderate, but appeared to be robust even with new datasets from centers not included in training the automated methods.