Author Of 1 Presentation
FC03.02 - A step forward toward the fully automated assessment of the central vein sign
Abstract
Background
A deep-learning prototype method, called CVSNet, was recently introduced for the automated detection of the central vein sign (CVS) in brain lesions and demonstrated effective and accurate discrimination of multiple sclerosis (MS) from its mimics. However, this method solely considered focal lesions displaying the central vein sign (CVS+) or not (CVS−), therefore requiring a manual pre-selection of the lesions to be evaluated by eliminating the so-called excluded lesions (CVSe) as defined by the NAIMS criteria. CVSe lesions may however play an important role in differential diagnosis. Moreover, extending the automated CVS classification to these lesions would facilitate the integration of CVSNet with existing MS lesion segmentation algorithms in a fully automated pipeline.
Objectives
To develop an improved version of the CVSNet prototype method able to classify all types of lesions (CVS+, CVS− and CVSe).
Methods
Patients with an established MS or CIS diagnosis (RRMS 29; SPMS 10; PPMS 10; CIS 1; mean ± SD age: 50 ± 11 years; male/female: 23/27), and healthy controls (n=8; mean ± SD age: 41 ± 9 years; male/female: 5/3), underwent 3T brain MRI (MAGNETOM Skyra and MAGNETOM Prisma, Siemens Healthcare, Erlangen, Germany, or Achieva, Philips Healthcare, Best, Netherlands). Brain lesions were automatically segmented and manually corrected by a single rater. CVS assessment was conducted on FLAIR* images by two raters, according to the NAIMS guidelines, yielding 1542 CVS+, 1004 CVS−, and 1131 CVSe lesions. A convolutional neural network (CNN) based on the CVSnet architecture was trained with different configurations using 3021 samples (1261 CVS+, 847 CVS−, and 913 CVSe) and evaluated in 656 unseen samples (281 CVS+, 157 CVS−, and 218 CVSe, from 13 patients) for final testing. The configurations relied on different combinations of the following channels as input: (i) FLAIR*, (ii) T2*, (iii) lesion mask, and (iv) CSF and brain tissue concentration maps obtained from a partial-volume estimation algorithm. Lesion-wise classification performance was evaluated for the different configurations by estimating the sensitivity, specificity, and accuracy for each lesion class.
Results
The results were similar across the different configurations. The best performance in the unseen testing set was obtained when all channels were used as input (sensitivity: 0.71, 0.73; specificity: 0.71, 0.81; and accuracy: 0.71, 0.79 for CVS+, CVS−, respectively). For CVSe, this approach achieved 0.52 sensitivity, 0.94 specificity, and 0.80 accuracy.
Conclusions
We introduced a modified CVSNet prototype method that can analyze the presence of the central vein for all types of brain lesions, enabling its integration with current MS lesion segmentation algorithms. This new feature will allow a fully automated assessment of the CVS in patients’ brains, speeding up the evaluation of CVS as a diagnostic biomarker for differentiating MS from mimicking diseases.
Author Of 1 Presentation
P0562 - Cortical lesions are not associated with leptomeningeal enhancement in a cohort of adults with MS. (ID 420)
Abstract
Background
Focal leptomeningeal enhancement (LME) on MRI is more commonly seen in neuroinflammatory diseases than in noninflammatory neurological diseases or healthy controls. In MS, meningeal inflammatory infiltrates sometimes overlie cortical demyelination in pathological samples, but studies linking cortical lesions to LME have had equivocal results to date.
Objectives
To evaluate the association between LME and cortical lesions in vivo and to assess the relationship between LME number and disease severity.
Methods
59 adults with MS (40 with relapsing remitting MS and 19 with primary or secondary progressive MS) underwent clinical testing (Multiple Sclerosis Functional Composite), 3 tesla (3T) MRI with gadolinium, and 7T non-gadolinium MRI within 6 months of the 3T MRI. 7T T1w MP2RAGE and T2*w gradient-echo images (both 0.5mm isometric) were used to identify cortical lesions, which were classified as leukocortical, intracortical, or subpial. Foci of LME were identified using post-gadolinium T2-FLAIR images and post–pre T2-FLAIR subtraction images. The spatial relationship between LME and cortical lesions was investigated, as was the clinical relationship between LME number and disease severity.
Results
66% of individuals (39/59) had no LME. 20% (12/59) had 1 focus of LME, and 14% (8/59) had >1 focus of LME. Median cortical lesion number was 20 in people without LME (IQR 91, range 0–206), 18 in people with 1 LME (IQR 12, range 0–182, p>0.05 vs no LME), and 39 in people with >1 LME (IQR 64, range 4–133, p>0.05 vs no LME). There was no difference in leukocortical, intracortical, or subpial lesion number between people with 0, 1, or >1 LME (p>0.05). None of the identified foci of LME was adjacent to a cortical lesion, though 13% of LME foci (4/31) were situated in the same sulcus as a cortical lesion. Median expanded disability status scale (EDSS) was higher in people with >1 focus of LME (5.5, range 1–7.5) compared to people without LME (median 1.5, range 0–7, p<0.05). EDSS was correlated with total cortical lesion number (rs=0.507, p<0.0001, ß=0.024) and subpial lesion number (rs=0.462, p<0.001, ß=0.028).
Conclusions
There was no association between number of LME foci and number of total cortical lesions or any cortical lesion subtype in our data. This suggests that LME cannot be taken to indicate ongoing inflammation overlying cortical demyelination. Further studies are needed to determine the histopathological basis of focal LME in MS and its relation, if any, to prior leptomeningeal inflammation.