Exploratory Radiomic Analysis of Conventional vs. Quantitative Brain MRI: Toward Automatic Diagnosis of Early Multiple Sclerosis

Conventional magnetic resonance imaging (cMRI) is poorly sensitive to pathological changes related to multiple sclerosis (MS) in normal-appearing white matter (NAWM) and gray matter (GM), with the added difficulty of not being very reproducible. Quantitative MRI (qMRI), on the other hand, attempts to represent the physical properties of tissues, making it an ideal candidate for quantitative medical image analysis or radiomics. We therefore hypothesized that qMRI-based radiomic features have added diagnostic value in MS compared to cMRI. This study investigated the ability of cMRI (T1w) and qMRI features extracted from white matter (WM), NAWM, and GM to distinguish between MS patients (MSP) and healthy control subjects (HCS). We developed exploratory radiomic classification models on a dataset comprising 36 MSP and 36 HCS recruited in CHU Liege, Belgium, acquired with cMRI and qMRI. For each image type and region of interest, qMRI radiomic models for MS diagnosis were developed on a training subset and validated on a testing subset. Radiomic models based on cMRI were developed on the entire training dataset and externally validated on open-source datasets with 167 HCS and 10 MSP. Ranked by region of interest, the best diagnostic performance was achieved in the whole WM. Here the model based on magnetization transfer imaging (a type of qMRI) features yielded a median area under the receiver operating characteristic curve (AUC) of 1.00 in the testing sub-cohort. Ranked by image type, the best performance was achieved by the magnetization transfer models, with median AUCs of 0.79 (0.69–0.90, 90% CI) in NAWM and 0.81 (0.71–0.90) in GM. The external validation of the T1w models yielded an AUC of 0.78 (0.47–1.00) in the whole WM, demonstrating a large 95% CI and a low sensitivity of 0.30 (0.10–0.70). This exploratory study indicates that qMRI radiomics could provide efficient diagnostic information using NAWM and GM analysis in MSP. T1w radiomics could be useful for a fast and automated check of conventional MRI for WM abnormalities once acquisition and reconstruction heterogeneities have been overcome. Further prospective validation is needed, involving more data for better interpretation and generalization of the results.

Download Full-text

Quantitative MRI Demonstrates Abnormality of the Fornix and Cingulum in Multiple Sclerosis

Multiple Sclerosis International ◽

10.1155/2013/838719 ◽

2013 ◽

Vol 2013 ◽

pp. 1-9 ◽

Cited By ~ 8

Author(s):

Stephanie B. Syc ◽

Daniel M. Harrison ◽

Shiv Saidha ◽

Michaela Seigo ◽

Peter A. Calabresi ◽

...

Keyword(s):

Multiple Sclerosis ◽

White Matter ◽

Healthy Volunteers ◽

Cognitive Dysfunction ◽

Diffusion Tensor ◽

Magnetization Transfer ◽

Mean Diffusivity ◽

Brain Structures ◽

Quantitative Mri ◽

Magnetization Transfer Ratio

Objective. To characterize MR signal changes associated with tissue damage in the fornix and cingulum in multiple sclerosis (MS) using quantitative MRI measures and to determine associations with cognitive dysfunction.Background. The fornix and cingulum are white-matter bundles that carry information related to cognition. While cognitive dysfunction is reported in 40–60% of MS patients, the neuroanatomical correlates of cognitive impairment remain incompletely understood.Methods. The cingulum, pillars of the fornix, and corticospinal tract were segmented by fiber tracking via diffusion tensor imaging. Average tract-specific fractional anisotropy (FA), mean diffusivity (MD), and magnetization transfer ratio (MTR) were compared in MS cases and healthy volunteers. Associations with clinical measures and neuropsychological tests were derived by multivariate linear regression.Results. Fornix FA (P=0.004) and MTR (P=0.005) were decreased, and fornix MD (P<0.001) and cingulum MD (P<0.001) increased, in MS cases (n=101) relative to healthy volunteers (n=16) after adjustment for age and sex. Lower fornix FA and MTR, and higher fornix MD andλ∥, were correlated with lower PASAT-3 scores, but not with slower 25FTW times. Lower PASAT-3 scores were associated with lower cingulum FA and higher MD andλ⊥.Conclusions. Cognitive dysfunction in MS may involve damage to a widespread network of brain structures, including white-matter pathways within the limbic system.

Download Full-text

Selective inversion recovery quantitative magnetization transfer imaging: Toward a 3 T clinical application in multiple sclerosis

Multiple Sclerosis Journal ◽

10.1177/1352458519833018 ◽

2019 ◽

Vol 26 (4) ◽

pp. 457-467 ◽

Cited By ~ 1

Author(s):

Francesca Bagnato ◽

Giulia Franco ◽

Fei Ye ◽

Run Fan ◽

Patricia Commiskey ◽

...

Keyword(s):

Multiple Sclerosis ◽

White Matter ◽

Magnetization Transfer ◽

Brain Mri ◽

Inversion Recovery ◽

Clinical Disease ◽

Normal Appearing White Matter ◽

Selective Inversion ◽

Pool Model ◽

Magnetic Resonance Imaging Mri

Background:Assessing the degree of myelin injury in patients with multiple sclerosis (MS) is challenging due to the lack of magnetic resonance imaging (MRI) methods specific to myelin quantity. By measuring distinct tissue parameters from a two-pool model of the magnetization transfer (MT) effect, quantitative magnetization transfer (qMT) may yield these indices. However, due to long scan times, qMT has not been translated clinically.Objectives:We aim to assess the clinical feasibility of a recently optimized selective inversion recovery (SIR) qMT and to test the hypothesis that SIR-qMT-derived metrics are informative of radiological and clinical disease-related changes in MS.Methods:A total of 18 MS patients and 9 age- and sex-matched healthy controls (HCs) underwent a 3.0 Tesla (3 T) brain MRI, including clinical scans and an optimized SIR-qMT protocol. Four subjects were re-scanned at a 2-week interval to determine inter-scan variability.Results:SIR-qMT measures differed between lesional and non-lesional tissue ( p < 0.0001) and between normal-appearing white matter (NAWM) of patients with more advanced disability and normal white matter (WM) of HCs ( p < 0.05). SIR-qMT measures were associated with lesion volumes, disease duration, and disability scores ( p ⩽ 0.002).Conclusion:SIR-qMT at 3 T is clinically feasible and predicts both radiological and clinical disease severity in MS.

Download Full-text

Characterization of white matter damage in animal models of multiple sclerosis by magnetization transfer ratio and quantitative mapping of the apparent bound proton fraction f*

Multiple Sclerosis Journal ◽

10.1177/1352458508096006 ◽

2009 ◽

Vol 15 (1) ◽

pp. 16-27 ◽

Cited By ~ 29

Author(s):

M Rausch ◽

PS Tofts ◽

P Lervik ◽

AR Walmsley ◽

A Mir ◽

...

Keyword(s):

Multiple Sclerosis ◽

White Matter ◽

Animal Models ◽

Tissue Damage ◽

Magnetization Transfer ◽

Transverse Relaxation Time ◽

Magnetization Transfer Ratio ◽

White Matter Damage ◽

Transfer Ratio ◽

Quantitative Mapping

Quantitative magnetization transfer magnetic resonance imaging (qMT-MRI) can be used to improve detection of white matter tissue damage in multiple sclerosis (MS) and animal models thereof. To study the correlation between MT parameters and tissue damage, the magnetization transfer ratio (MTR), the parameter f* (closely related to the bound proton fraction) and the bound proton transverse relaxation time T2B of lesions in a model of focal experimental autoimmune encephalomyelitis (EAE) were measured on a 7T animal scanner and data were compared with histological markers indicative for demyelination, axonal density, and tissue damage. A clear spatial correspondence was observed between reduced values of MTR and demyelination in this animal model. We observed two different levels of MTR and f* reduction for these lesions. One was characterized by a pronounced demyelination and the other corresponded to a more severe loss of the cellular matrix. Changes in f* were generally more pronounced than those of MTR in areas of demyelination. Moreover, a reduction of f* was already observed for tissue where MTR was virtually normal. No changes in T2B were observed for the lesions. We conclude that MTR and qMT mapping are efficient and reliable readouts for studying demyelination in animal models of MS, and that the analysis of regional f* might be even superior to the analysis of MTR values. Therefore, quantitative mapping of f* from human brains might also improve the detection of white matter damage in MS.

Download Full-text

Gray and White Matter Demyelination and Remyelination Detected with Multimodal Quantitative MRI Analysis at 11.7T in a Chronic Mouse Model of Multiple Sclerosis

Frontiers in Neuroscience ◽

10.3389/fnins.2016.00491 ◽

2016 ◽

Vol 10 ◽

Cited By ~ 7

Author(s):

Alexandra Petiet ◽

Marie-Stéphane Aigrot ◽

Bruno Stankoff

Keyword(s):

Multiple Sclerosis ◽

White Matter ◽

Mouse Model ◽

Quantitative Mri

Download Full-text

Magnetization transfer and adiabatic T1ρ MRI reveal abnormalities in normal-appearing white matter of subjects with multiple sclerosis

Multiple Sclerosis Journal ◽

10.1177/1352458513515084 ◽

2013 ◽

Vol 20 (8) ◽

pp. 1066-1073 ◽

Cited By ~ 19

Author(s):

Silvia Mangia ◽

Adam F Carpenter ◽

Andy E Tyan ◽

Lynn E Eberly ◽

Michael Garwood ◽

...

Keyword(s):

Multiple Sclerosis ◽

White Matter ◽

Magnetization Transfer ◽

Magnetic Resonance Images ◽

Relaxation Methods ◽

Normal Appearing White Matter ◽

Relaxation Parameters ◽

Relapsing Remitting ◽

Cortical Gray Matter ◽

The Brain

Background: Diffuse abnormalities are known to occur within the brain tissue of multiple sclerosis (MS) patients that is “normal appearing” on T1-weighted and T2-weighted magnetic resonance images. Objectives: With the goal of exploring the sensitivity of novel MRI parameters to detect such abnormalities, we implemented an inversion-prepared magnetization transfer (MT) protocol and adiabatic T1ρ and T2ρ rotating frame relaxation methods. Methods: Nine relapsing–remitting MS patients and seven healthy controls were recruited. Relaxation parameters were measured in a single slice just above the lateral ventricles and approximately parallel to the AC-PC line. Results: The MT ratio of regions encompassing the normal-appearing white matter (NAWM) was different in MS patients as compared with controls ( p = 0.043); however, the T1 measured during off-resonance irradiation (T1sat) was substantially more sensitive than the MT ratio for detecting differences between groups ( p = 0.0006). Adiabatic T1ρ was significantly prolonged in the NAWM of MS patents as compared to controls (by 6%, p = 0.026), while no differences were found among groups for T2ρ. No differences among groups were observed in the cortical gray matter for any relaxation parameter. Conclusions: The results suggest degenerative processes occurring in the NAWM of MS, likely not accompanied by significant abnormalities in iron content.

Download Full-text

Focal multiple sclerosis lesions abound in ‘normal appearing white matter’

Multiple Sclerosis Journal ◽

10.1177/1352458511415305 ◽

2011 ◽

Vol 17 (11) ◽

pp. 1313-1323 ◽

Cited By ~ 33

Author(s):

Niraj Mistry ◽

Emma C Tallantyre ◽

Jennifer E Dixon ◽

Nicolas Galazis ◽

Tim Jaspan ◽

...

Keyword(s):

Multiple Sclerosis ◽

White Matter ◽

High Field ◽

White Matter Lesions ◽

7 Tesla ◽

Ultra High Field ◽

Rapid Acquisition ◽

Normal Appearing White Matter ◽

Mri Scans ◽

Image Type

Background: The ‘normal appearing white matter’ (NAWM) in multiple sclerosis (MS) is known to be abnormal using quantitative magnetic resonance (MR) techniques. The aetiology of the changes in NAWM remains debatable. Objective: To investigate whether high-field and ultra high-field T1-weighted magnetization prepared rapid acquisition gradient echo (MPRAGE) MRI enables detection of MS white matter lesions in areas defined as NAWM using high-field T2-weighted fluid attenuation inversion recovery (FLAIR) MRI; that is, to ascertain whether undetected lesions are likely contributors to the burden of abnormality in similarly defined NAWM. Methods: Fourteen MS patients underwent MRI scans using 3T FLAIR and MPRAGE and 7 Tesla (7T) MPRAGE sequences. Independent observers identified lesions on 3T FLAIR and (7T and 3T) MPRAGE images. The detection of every individual lesion was then compared for each image type. Results: We identified a total of 812 white matter lesions on 3T FLAIR. Using 3T MPRAGE, 186 additional lesions were detected that were not detected using 3T FLAIR. Using 7T MPRAGE, 231 additional lesions were detected that were not detected using 3T FLAIR. Conclusions: MRI with 3T and 7T MPRAGE enables detection of MS lesions in areas defined as NAWM using 3T FLAIR. Focal MS lesions contribute to the abnormalities known to exist in the NAWM.

Download Full-text

Normal-Appearing White Matter Changes in Multiple Sclerosis: Magnetization Transfer

Normal-appearing White and Grey Matter Damage in Multiple Sclerosis ◽

10.1007/978-88-470-2127-3_4 ◽

2004 ◽

pp. 37-45

Author(s):

F. Fazekas ◽

C. Enzinger ◽

S. Ropele

Keyword(s):

Multiple Sclerosis ◽

White Matter ◽

Magnetization Transfer ◽

White Matter Changes ◽

Normal Appearing White Matter

Download Full-text

Surface-in pathology in multiple sclerosis: a new view on pathogenesis?

Brain ◽

10.1093/brain/awab025 ◽

2021 ◽

Author(s):

Matteo Pardini ◽

J William L Brown ◽

Roberta Magliozzi ◽

Richard Reynolds ◽

Declan T Chard

Keyword(s):

Multiple Sclerosis ◽

White Matter ◽

Grey Matter ◽

Magnetization Transfer ◽

Neuronal Loss ◽

Magnetization Transfer Ratio ◽

Supporting Evidence ◽

Transfer Ratio ◽

Meningeal Inflammation

Abstract While multiple sclerosis can affect any part of the CNS, it does not do so evenly. In white matter it has long been recognized that lesions tend to occur around the ventricles, and grey matter lesions mainly accrue in the outermost (subpial) cortex. In cortical grey matter, neuronal loss is greater in the outermost layers. This cortical gradient has been replicated in vivo with magnetization transfer ratio and similar gradients in grey and white matter magnetization transfer ratio are seen around the ventricles, with the most severe abnormalities abutting the ventricular surface. The cause of these gradients remains uncertain, though soluble factors released from meningeal inflammation into the CSF has the most supporting evidence. In this Update, we review this ‘surface-in’ spatial distribution of multiple sclerosis abnormalities and consider the implications for understanding pathogenic mechanisms and treatments designed to slow or stop them.

Download Full-text