Author Of 1 Presentation
PS16.05 - Application of deep-learning to NMOSD and unclassified seronegative patients
Abstract
Background
Current diagnostic criteria of neuromyelitis optica spectrum disorders (NMOSD) allow the diagnosis of aquaporin-4 (AQP4) seropositive patients with limited manifestations, whereas seronegative patients with limited phenotypes remain unclassified and are usually considered as prodromal phases of multiple sclerosis (MS) or different entities themselves. Nowadays, there is great effort to perform an automatic diagnosis of different neurological diseases using deep-learning-based imaging diagnostics, which is a form of artificial intelligence, allowing predicting or making decisions without a priori human intervention.
Objectives
To provide a deep-learning classification of NMOSD patients with different serological profiles and to compare these results with their clinical evolution.
Methods
228 T2- and T1-weighted brain MRIs were acquired from patients with AQP4-seropositive NMOSD (n=85), early MS (n=95), AQP4-seronegative NMOSD (n=11, 3 with anti-myelin oligodendrocyte glycoprotein antibodies) and unclassified double-seronegative limited phenotypes (n=17 idiopathic recurrent optic neuritis [IRON], n=20 idiopathic recurrent myelitis [IRM]). The latter had a clinical re-evaluation after 4-year follow-up. The neural network architecture was based on four 3D convolutional layers. It was trained and validated on MRI scans (n=180) from AQP4-seropositive NMOSD and MS patients. Then, it was applied to AQP4-seronegative NMOSD and double-seronegative patients with limited phenotypes to evaluate their classification as NMOSD or MS in comparison with their clinical follow-up.
Results
The final algorithm discriminated between AQP-4-seropositive NMOSD and MS with an accuracy of 0.95. Forty-seven/48 (97.9%) seronegative patients were classified as NMOSD (one patient with IRON was classified as MS). Clinical follow-up was available in 27/37 (73%) double-seronegative limited phenotypes: one patient evolved to MS, three developed NMOSD and the others did not change phenotype.
Conclusions
Deep-learning may help in the diagnostic work-up of NMOSD. Our findings support the inclusion of AQP4-seronegative patients to the spectrum of NMO and suggest its enlargement to double-seronegative limited phenotypes.
Author Of 2 Presentations
P0247 - Comparison of the 2017 and 2010 revisions of the McDonald criteria in patients with cis suggestive of MS: a multicentre MAGNIMS study (ID 1121)
- P. Preziosa
- A. Meani
- A. Rovira
- S. Mesaros
- O. Ciccarelli
- J. Frederiksen
- C. Enzinger
- F. Barkhof
- C. Gasperini
- E. Fainardi
- W. Brownlee
- J. Drulovic
- X. Montalban
- S. Cramer
- M. Kalil
- M. Hagens
- S. Ruggieri
- G. Dalla Costa
- V. Martinelli
- K. Miszkiel
- M. Tintore
- G. Comi
- I. Dekker
- B. Uitdehaag
- J. Ivanović
- M. Amato
- M. Rocca
Abstract
Background
In 2017, a revision of the 2010 McDonald criteria for multiple sclerosis (MS) diagnosis in clinically isolated syndrome (CIS) patients has been proposed. However, its validation in a large multicenter cohort of CIS patients is still needed.
Objectives
To compare the performance of 2017 and 2010 revisions of the McDonald criteria with respect to MS development in a large multicentric cohort of CIS suggestive of MS.
Methods
Brain and spinal cord magnetic resonance imaging (MRI) and cerebrospinal fluid (CSF) examination obtained ≤5 months from CIS onset and a follow-up brain MRI acquired ≤15 months from CIS onset were assessed in 626 CIS patients from 9 European MS centres. The occurrence of a second clinical attack (clinically definite [CD] MS) was recorded. Performances of the 2017 and 2010 revisions of McDonald criteria for dissemination in space (DIS), time (DIT) and DIS plus DIT, also including OCB assessment, were evaluated with a time-dependent receiver operating characteristic curve analysis. Median time to MS diagnosis for the different sets of criteria was estimated through Kaplan-Meier curves.
Results
At the last evaluation (median=61.9 months [IQR=39.1-102.5]), 319 (51%) of 626 patients had CDMS. At 36 months, for DIS, the 2017 MRI criteria had higher sensitivity (0.84 [95% CI=0.79-0.88] vs 0.77 [0.72-0.82]), lower specificity (0.33 [0.28-0.39] vs 0.40 [0.35-0.46]), and similar area under the curve values (AUC, 0.59 [0.55-0.62] for both). The 2017 DIS plus DIT MRI criteria had higher sensitivity (0.68 [0.63-0.74] vs 0.62 [0.56-0.68]), lower specificity (0.55 [0.49-0.61] vs 0.62 [0.56-0.68]), and similar AUC values (0.62 [0.58-0.66] for both). CSF-specific OCB assessment as part of the 2017 criteria revision, increased the sensitivity (0.81 [0.75-0.85]), decreased specificity (0.40 [0.34-0.46]) and preserved AUC values (0.60 [0.56-0.64]). Median time to MS diagnosis was earlier with the 2017 revision compared to the 2010 or CDMS criteria, especially with OCB assessment (2017 revision with OCBs=3.6 months [3.1-4.0], 2017 revision without OCB=11.6 months [7.8-13.5], 2010 revision=13.9 months [12.4-15.3], CDMS=56.3 months [43.8-76.0]).
Conclusions
The 2017 revision of the McDonald criteria showed overall similar accuracy to the 2010 McDonald criteria in predicting CDMS development. The suggested modifications are expected to simplify the clinical use of MRI criteria without reducing accuracy and allow an earlier diagnosis of MS.
P0982 - MR T2-relaxation time as an indirect measure of brain water accumulation in Neuromyelitis Optica Spectrum Disorders (ID 1077)
Abstract
Background
One of the main unsolved issues in the clinical management of neuromyelitis optica spectrum disorders (NMOSD) is the lack of biomarkers predicting short-term relapses. In physiological conditions, the blood brain barrier (BBB) protects the CNS from water unbalance, with aquaporin-4 (AQP4) water channels on astrocytes podocytes being the main regulator of water influx and efflux. In NMOSD, BBB integrity might be threatened by the presence of antibodies targeting AQP4 water channels and triggering complement-mediated astrocytes damage. In line with this, increased T2-signal in acute lesions (“bright spotty lesions”) is considered specific for NMOSD. However, it remains unexplored whether these patients present a chronic water unbalance.
Objectives
To provide an indirect estimation of brain water content in NMOSD by measuring T2-relaxation time (T2rt) and to assess whether it differs in patients having a short-term relapse.
Methods
In this multicenter MR study, T2rt was calculated from brain dual echo turbo spin echo images assuming a mono exponential decay. T2rt maps of normal appearing white matter (NAWM), gray matter (GM) and basal ganglia were obtained from 77 AQP4-positive NMOSD and 84 HC. Short-term relapses were defined as those occurring within one month before or after MRI scan. Differences between NMOSD and HC were assessed with age-, sex- and site-adjusted linear models. ROC analyses were run to identify discriminators between stable and short-term relapsing patients.
Results
NMOSD patients and HC had similar ages. Compared to HC, T2rt was increased in the GM (103 vs 97 ms), NAWM (88 vs 84 ms) and putamen (75 vs 72 ms) of NMOSD patients (p<0.001 for all). Short-term relapses occurred in 20/77 (26%) of patients. According to ROC analysis, T2rt cut-offs of 87 ms in the NAWM, 87 ms in the thalamus and 88 ms in the caudatus were able to discriminate between short-term relapsing and stable patients with good accuracy (AUC=0.70, 0.76 and 0.79 respectively, p≤ 0.027).
Conclusions
NMOSD patients had increased T2rt values, in line with the hypothesis of subclinical water accumulation in this disorder. The burden of T2rt alterations might be useful for identifying those patients with incipient or recent relapses.