Three-Dimensional Shape and Surface Features Distinguish Multiple Sclerosis Lesions from Nonspecific White Matter Disease

Differentiating multiple sclerosis from non-specific white matter changes using a convolutional neural network image classification model

BACKGROUND

The diagnosis of multiple sclerosis (MS) relies heavily on neuroimaging with magnetic resonance imaging (MRI) and exclusion of mimics. This can be a challenging task due to radiological overlap in several disorders and may require ancillary testing or longitudinal follow up. One of the most common radiological MS mimickers is non-specific white matter disease (NSWMD). We aimed to develop and evaluate models leveraging machine learning algorithms to help distinguish MS and NSWMD.

METHODS

All adult patients who underwent MRI brain using a demyelinating protocol with available electronic medical records between 2015 and 2019 at Cleveland Clinic affiliated facilities were included. Diagnosis of MS and NSWMD were assessed from clinical documentation. Those with a diagnosis of MS and NSWMD were matched using total T2 lesion volume (T2LV) and used to train models with logistic regression and convolutional neural networks (CNN). Performance metrices were reported for each model.

RESULTS

A total of 250 NSWMD MRI scans were identified, and 250 unique MS MRI scans were matched on T2LV. Cross validated logistic regression model was able to use 20 variables (including spinal cord area, regional volumes, and fractions) to predict MS compared to NSWMD with 68.0% accuracy while the CNN model was able to classify MS compared to NSWMD in two independent validation and testing cohorts with 77% and 78% accuracy on average.

CONCLUSION

Automated methods can be used to differentiate MS compared to NSWMD. These methods can be used to supplement currently available diagnostic tools for patients being evaluated for MS.

Neuroimaging to monitor worsening of multiple sclerosis: advances supported by the grant for multiple sclerosis innovation

Key unmet needs in multiple sclerosis (MS) include detection of early pathology, disability worsening independent of relapses, and accurate monitoring of treatment response. Collaborative approaches to address these unmet needs have been driven in part by industry-academic networks and initiatives such as the Grant for Multiple Sclerosis Innovation (GMSI) and Multiple Sclerosis Leadership and Innovation Network (MS-LINK™) programs. We review the application of recent advances, supported by the GMSI and MS-LINK™ programs, in neuroimaging technology to quantify pathology related to central pathology and disease worsening, and potential for their translation into clinical practice/trials. GMSI-supported advances in neuroimaging methods and biomarkers include developments in magnetic resonance imaging, positron emission tomography, ocular imaging, and machine learning. However, longitudinal studies are required to facilitate translation of these measures to the clinic and to justify their inclusion as endpoints in clinical trials of new therapeutics for MS. Novel neuroimaging measures and other biomarkers, combined with artificial intelligence, may enable accurate prediction and monitoring of MS worsening in the clinic, and may also be used as endpoints in clinical trials of new therapies for MS targeting relapse-independent disease pathology.

Selective vulnerability of brainstem and cervical spinal cord regions in people with non-progressive multiple sclerosis of Black or African American and European ancestry

Background:

We evaluated imaging features suggestive of neurodegeneration within the brainstem and upper cervical spinal cord (UCSC) in non-progressive multiple sclerosis (MS).

Methods:

Standardized 3-Tesla three-dimensional brain magnetic resonance imaging (MRI) studies were prospectively acquired. Rates of change in volume, surface texture, curvature were quantified at the pons and medulla-UCSC. Whole and regional brain volumes and T2-weighted lesion volumes were also quantified. Independent regression models were constructed to evaluate differences between those of Black or African ancestry (B/AA) and European ancestry (EA) with non-progressive MS.

Results:

209 people with MS (pwMS) having at least two MRI studies, 29% possessing 3–6 timepoints, resulted in 487 scans for analysis. Median follow-up time between MRI timepoints was 1.33 (25th–75th percentile: 0.51–1.98) years. Of 183 non-progressive pwMS, 88 and 95 self-reported being B/AA and EA, respectively. Non-progressive pwMS demonstrated greater rates of decline in pontine volume ( p < 0.0001) in B/AA and in medulla-UCSC volume ( p < 0.0001) for EA pwMS. Longitudinal surface texture and curvature changes suggesting reduced tissue integrity were observed at the ventral medulla-UCSC ( p < 0.001), dorsal pons ( p < 0.0001) and dorsal medulla ( p < 0.0001) but not the ventral pons ( p = 0.92) between groups.

Conclusions:

Selectively vulnerable regions within the brainstem-UCSC may allow for more personalized approaches to disease surveillance and management.

Brain white matter hyperintensity lesion characterization in 3D T2 fluid-attenuated inversion recovery magnetic resonance images: Shape, texture, and their correlations with potential growth

Analyses of age-related white matter hyperintensity (WMH) lesions manifested in T2 fluid-attenuated inversion recovery (FLAIR) magnetic resonance images (MRI) have been mostly on understanding the size and location of the WMH lesions and rarely on the morphological characterization of the lesions. This work extends our prior analyses of the morphological characteristics and texture of WMH from 2D to 3D based on 3D T2 FLAIR images. 3D Zernike transformation was used to characterize WMH shape; a fuzzy logic method was used to characterize the lesion texture. We then clustered 3D WMH lesions into groups based on their 3D shape and texture features. A potential growth index (PGI) to assess dynamic changes in WMH lesions was developed based on the image texture features of the WMH lesion penumbra. WMH lesions with various sizes were segmented from brain images of 32 cognitively normal older adults. The WMH lesions were divided into two groups based on their size. Analyses of Variance (ANOVAs) showed significant differences in PGI among WMH shape clusters (P = 1.57 × 10-3 for small lesions; P = 3.14 × 10-2 for large lesions). Significant differences in PGI were also found among WMH texture group clusters (P = 1.79 × 10-6). In conclusion, we presented a novel approach to characterize the morphology of 3D WMH lesions and explored the potential to assess the dynamic morphological changes of WMH lesions using PGI.

Neuroradiological differentiation of white matter lesions in patients with multiple sclerosis and Fabry disease

Objective

White matter lesions (WML) in multiple sclerosis (MS) differ from vascular WML caused by Fabry disease (FD). However, in atypical cases the discrimination can be difficult and may vary between individual raters. The aim of this study was to evaluate interrater reliability of WML differentiation between MS and FD patients.

Materials and methods

Brain MRI scans of 21 patients with genetically confirmed FD were compared to 21 matched patients with MS. Pseudonymized axial FLAIR sequences were assessed by 6 blinded raters and attributed to either the MS or the FD group to investigate interrater reliability. Additionally, localization of WML was compared between the two groups.

Results

The median age of patients was 46 years (IQR 35–58). Interrater reliability was moderate with a Fleiss' Kappa of 0.45 (95%CI 0.3–0.59). Overall, 85% of all ratings in the MS group and 75% in the FD group were correct. However, only 38% of patients with MS and 33% of patients with FD were correctly identified by all 6 raters. WML involving the corpus callosum (p < 0.001) as well as juxtacortical (p < 0.001) and infratentorial lesions (p = 0.03) were more frequently observed in MS patients.

Conclusion

Interrater reliability regarding visual differentiation of WML in MS from vascular WML in FD on standard axial FLAIR images alone is only moderate, despite the distinctive features of lesions in each group.

Utility of shape evolution and displacement in the classification of chronic multiple sclerosis lesions

The accurate recognition of multiple sclerosis (MS) lesions is challenged by the high sensitivity and imperfect specificity of MRI. To examine whether longitudinal changes in volume, surface area, 3-dimensional (3D) displacement (i.e. change in lesion position), and 3D deformation (i.e. change in lesion shape) could inform on the origin of supratentorial brain lesions, we prospectively enrolled 23 patients with MS and 11 patients with small vessel disease (SVD) and performed standardized 3-T 3D brain MRI studies. Bayesian linear mixed effects regression models were constructed to evaluate associations between changes in lesion morphology and disease state. A total of 248 MS and 157 SVD lesions were studied. Individual MS lesions demonstrated significant decreases in volume < 3.75mm³ (p = 0.04), greater shifts in 3D displacement by 23.4% with increasing duration between MRI time points (p = 0.007), and greater transitions to a more non-spherical shape (p < 0.0001). If 62.2% of lesions within a given MRI study had a calculated theoretical radius > 2.49 based on deviation from a perfect 3D sphere, a 92.7% in-sample and 91.2% out-of-sample accuracy was identified for the diagnosis of MS. Longitudinal 3D shape evolution and displacement characteristics may improve lesion classification, adding to MRI techniques aimed at improving lesion specificity.

Overview of the use of 3D printing in forensic medicine

In forensic medicine the use of so-called 3D printing is a niche application, whereas developments elsewhere in this field are rapidly advancing worldwide. The most common and widespread technology is fusion deposit modelling with polylactic acids (PLA). Although the equipment and materials may be relatively inexpensive and 3D printing relatively fast, the resulting end products tend to also have negative properties, such as poor durability and mechanical anisotropy, which may be an issue depending on the application. In forensic medicine, applications in the field of weapons technology and biomechanical models are realistic and 3D printing is already being used for demonstrations at court hearings and in teaching and also as a technique for building spare parts or accessories. Having a low-cost option for rapid prototyping on-site is particularly useful for the development phase. For finished 3D designs more expensive manufacturing options with a choice of materials with significantly broader mechanical or thermal properties are available. As the technology is undergoing major changes, one should carefully consider whether to enter the field oneself, buy own hardware, use a 3D printing service or seek cooperation possibly with a nearby partner.

African Americans experience disproportionate neurodegenerative changes in the medulla and upper cervical spinal cord in early multiple sclerosis

OBJECTIVE

To compare the temporal changes in the 3-dimensional (3D) structure of the medulla-upper cervical spinal cord region in African American (AA) and white multiple sclerosis (MS) patients to identify early patterns of anatomical change prior to progressive symptom development.

METHODS

Standardized 3-Tesla 3D brain MRI studies were performed at two time points on AA and white MS patients along with controls. Longitudinal changes in volume, surface area, tissue compliance, and surface texture measured in total and within ventral and dorsal compartments were studied. Independent regression models were constructed to evaluate differences between groups.

RESULTS

Thirty-five individuals were studied, 10 AA with MS (female (F): 8; median age [IQR]=33.8 years (y) [10.9], median disease duration: 11.8y [11.3]), 20 white MS patients (F: 10; 35.6y [17.4], 7.23y [8.83], and 5 controls (F: 2, 51.8y [10.2]). Expanded Disability Status Scale scores were 0.0 at baseline and at the second MRI time point. Within the medulla-upper cervical spinal cord, AA versus white MS patients exhibited greater rates of atrophy in total (p<0.0001) and within the ventral (p<0.0001) and dorsal (p<0.0001) compartments, reduced surface area (p<0.0001), and reduced tissue compliance in the ventral (p=0.002) and dorsal (p=0.0005) compartments. The rate of change at the dorsal surface, but not the ventral surface, between MRI time points was also greater in AA relative to white MS patients (p<0.0001).

CONCLUSION

Structural changes in distinct anatomical regions of the medulla-upper cervical spinal cord may be reflective of early and disproportionate neurodegeneration in AA MS as compared to whites.

MRI-based prediction of conversion from clinically isolated syndrome to clinically definite multiple sclerosis using SVM and lesion geometry

Neuroanatomical pattern classification using support vector machines (SVMs) has shown promising results in classifying Multiple Sclerosis (MS) patients based on individual structural magnetic resonance images (MRI). To determine whether pattern classification using SVMs facilitates predicting conversion to clinically definite multiple sclerosis (CDMS) from clinically isolated syndrome (CIS). We used baseline MRI data from 364 patients with CIS, randomised to interferon beta-1b or placebo. Non-linear SVMs and 10-fold cross-validation were applied to predict converters/non-converters (175/189) at two years follow-up based on clinical and demographic data, lesion-specific quantitative geometric features and grey-matter-to-whole-brain volume ratios. We applied linear SVM analysis and leave-one-out cross-validation to subgroups of converters (n = 25) and non-converters (n = 44) based on cortical grey matter segmentations. Highest prediction accuracies of 70.4% (p = 8e-5) were reached with a combination of lesion-specific geometric (image-based) and demographic/clinical features. Cortical grey matter was informative for the placebo group (acc.: 64.6%, p = 0.002) but not for the interferon group. Classification based on demographic/clinical covariates only resulted in an accuracy of 56% (p = 0.05). Overall, lesion geometry was more informative in the interferon group, EDSS and sex were more important for the placebo cohort. Alongside standard demographic and clinical measures, both lesion geometry and grey matter based information can aid prediction of conversion to CDMS.

Precision medicine for multiple sclerosis promotes preventative medicine

Multiple sclerosis (MS) is a chronic, lifelong disease, currently without a cure that is responsible for significant neurological injury in young adults. Precision medicine for MS aims to provide a more exacting and refined approach toward management by providing recommendations based on disease subtype, clinical status, existing radiological data, para-clinical data, and other biological markers. To achieve better outcomes, the three stages of care-diagnosis, treatment, and management-should be optimized. However, as the temporal profile of disease behavior is highly variable in MS, and unlike outcomes from other chronic conditions (i.e., hypertension, diabetes mellitus, etc.), should precision medicine for MS be one that focuses more on disease prevention and lifestyle modifications beyond recommendations for the use of disease-modifying therapies? As scientific advancements continue within the field of neuroimmunology, and until reliable biomarkers that predict disease outcomes are available, success may be better achieved by focusing on modifiable factors to reduce future disability.

Brain White Matter Hyperintensity Lesion Characterization in T2 Fluid-Attenuated Inversion Recovery Magnetic Resonance Images: Shape, Texture, and Potential Growth

Prior methods in characterizing age-related white matter hyperintensity (WMH) lesions on T₂ fluid-attenuated inversion recovery (FLAIR) magnetic resonance images (MRI) have mainly been limited to understanding the sizes of, and occasionally the locations of WMH lesions. Systematic morphological characterization has been missing. In this work, we proposed innovative methods to fill this knowledge gap. We developed an innovative and proof-of-concept method to characterize and quantify the shape (based on Zernike transformation) and texture (based on fuzzy logic) of WMH lesions. We have also developed a multi-dimension feature vector approach to cluster WMH lesions into distinctive groups based on their shape and then texture features. We then developed an approach to calculate the potential growth index (PGI) of WMH lesions based on the image intensity distributions at the edge of the WMH lesions using a region-growing algorithm. High-quality T₂ FLAIR images containing clearly identifiable WMH lesions with various sizes from six cognitively normal older adults were used in our method development Analyses of Variance (ANOVAs) showed significant differences in PGI among WMH group clusters in terms of either the shape (P = 1.06 × 10⁻²) or the texture (P < 1 × 10⁻²⁰) features. In conclusion, we propose a systematic framework on which the shape and texture features of WMH lesions can be quantified and may be used to predict lesion growth in older adults.