CVSnet: A machine learning approach for automated central vein sign assessment in multiple sclerosis

Abstract

The central vein sign (CVS) is an efficient imaging biomarker for multiple sclerosis (MS) diagnosis, but its application in clinical routine is limited by inter-rater variability and the expenditure of time associated with manual assessment. We describe a deep learning-based prototype for automated assessment of the CVS in white matter MS lesions using data from three different imaging centers. We retrospectively analyzed data from 3 T magnetic resonance images acquired on four scanners from two different vendors, including adults with MS (n = 42), MS mimics (n = 33, encompassing 12 distinct neurological diseases mimicking MS) and uncertain diagnosis (n = 5). Brain white matter lesions were manually segmented on FLAIR* images. Perivenular assessment was performed according to consensus guidelines and used as ground truth, yielding 539 CVS-positive (CVS + ) and 448 CVS-negative (CVS − ) lesions. A 3D convolutional neural network ("CVSnet") was designed and trained on 47 datasets, keeping 33 for testing. FLAIR* lesion patches of CVS + /CVS − lesions were used for training and validation (n = 375/298) and for testing (n = 164/150). Performance was evaluated lesion-wise and subject-wise and compared with a state-of-the-art vesselness filtering approach through McNemar's test. The proposed CVSnet approached human performance, with lesion-wise median balanced accuracy of 81%, and subject-wise balanced accuracy of 89% on the validation set, and 91% on the test set. The process of CVS assessment, in previously manually segmented lesions, was~600-fold faster using the proposed CVSnet compared with human visual assessment (test set: 4 seconds vs. 40 minutes). On the validation and test sets, the lesion-wise performance outperformed the vesselness filter method (P < 0.001). The proposed deep learning prototype shows promising performance in differentiating MS from its mimics. Our approach was evaluated using data from different hospitals, enabling larger multicenter trials to evaluate the benefit of introducing the CVS marker into MS diagnostic criteria.Abbreviations used: AUC, area under the curve; CVS, central vein sign; FLAIR, 3D T2-weighted fluid-attenuated inversion recovery; GPU, graphics processing unit; MRI, magnetic resonance imaging; MS, multiple sclerosis; NIR, no information rate; ReLU, rectified linear unit; ROC, receiver operating characteristic; T2*-EPI, T2*-weighted segmented echo-planar imaging; 3D, threedimensional; WM, white matter.Pietro Maggi and Mário João Fartaria contributed equally to this work.

The central vein sign in multiple sclerosis patients with vascular comorbidities

“…Both lesion segmentation and CVS detection were done manually, limiting the applicability of this assessment in everyday clinical practice. Although automated methods for CVS detection are emerging, 14,37 In conclusion, older age and arterial HT significantly decrease the percentage of brain WM lesions featuring a visible central vein in MS. Although results of the CVS assessment should be interpreted with caution in these patients, the diagnostic yield of the previously proposed thresholds, 4 appears to be not affected.…”

“…Isotropic resolution of the 3D T2*-EPI was 0.55 mm 3 in Brussels and 0.65 mm 3 in Lausanne (Supplementary Table 1). 14…”

Paramagnetic Rim Lesions are Specific to Multiple Sclerosis: An International Multicenter 3T MRI Study

“…Also, it might not reflect real-world diagnosis in a newly presenting population. Although clinical implementation of these diagnostic biomarkers might be limited by the availability of 3T MR scanners, the preference for gadolinium injection to increase the sensitivity of CVS detection (but not necessarily of PRL detection), the current need for manual assessment of CVS and PRLs (although automatic detection algorithms 41,42 are being developed), and the lack of a product 3D T2*-EPI sequence on several scanner platforms, it is not difficult to envision these barriers being overcome.…”

RimNet: A deep 3D multimodal MRI architecture for paramagnetic rim lesion assessment in multiple sclerosis

“…Hence, image patches were extracted centered on the center of mass of the automatically detected lesions and linearly normalized between −1 and 1 ( Maggi et al, 2020 ). To ensure that the model was trained with reliable patches, we automatically removed lesions according to the following exclusion criteria.…”