An Automated Probabilistic Algorithm for the Detection of Central Vein Sign in Multiple Sclerosis

• Activity Number: 536 - Contributed Poster Presentations: Section on Statistics in Imaging
• Type: Contributed
• Date/Time: Wednesday, August 1, 2018 : 10:30 AM to 12:20 PM
• Sponsor: Section on Statistics in Imaging
• Author(s): Jordan Dworkin* and Pascal Sati and Andrew Solomon and Dzung Pham and Richard Watts and Melissa Martin and Daniel Ontaneda and Matthew K Schindler and Daniel S Reich and Russell T Shinohara
• Companies: University of Pennsylvania and National Institute of Neurological Disorders and Stroke and University of Vermont and Henry M. Jackson Foundation and University of Vermont and University of Pennsylvania and Cleveland Clinic and National Institute of Neurological Disorders and Stroke and National Institute of Neurological Disorders and Stroke and University of Pennsylvania
• Keywords: Neuroimaging; MRI; Multiple sclerosis

Abstract

Central vein sign (CVS) is a promising diagnostic biomarker for multiple sclerosis (MS). However, the utility of the CVS biomarker is limited by inter-rater differences in the adjudication of CVS, as well as the time burden required for the determination of CVS for each lesion in a patient's scan. The current study develops an automated technique for the detection of CVS in white matter lesions. The method is probabilistic, allows for site-specific segmentation methods, and has the potential to be robust to inter-site variability. The proposed algorithm is tested on 40 patients from the University of Vermont; 20 patients have MS, 20 patients do not. Using the automated technique, significant differences were found in the CVS biomarker between patients with MS (M = 0.55) and patients without MS (M = 0.31, p < 0.001). The algorithm was also found to show strong discriminative ability between MS and non-MS patients, with an AUC value of 0.88. The current study presents the first fully automated method for detecting central vessel sign in white matter lesions, and demonstrates its strong performance in a sample of MS and non-MS patients.