Research project

Diagnosing MS using Central Veins at Clinically Relevant Image Resolutions

Programme
STP
Specialty
Imaging Non-Ionising Radiation
Author
Edward Peake
Training location
Nottingham University Hospitals NHS Trust

Objective: To investigate the effect image resolution has on whether MS can be distinguished from MS-mimicking inflammatory conditions based on the appearance of central veins in white matter brain lesions.

Methods: This was a retrospective study of 49 patients with suspected MS, with 32 developing clinically definite MS. Subjects were imaged using 7T T2*-weighted MRI, and 3D interpolation was used to reduce the resolutions over a clinically relevant range. White matter lesions were identified and analysed for perivenous appearance according to consensus guidance. The frequency of perivenular lesions in MS versus MS-mimicking condition was assessed using the proportion of lesions with central veins found in other neurological conditions from previous studies.

Results: At 7T MRI, out of 1023 lesions identified in patients with MS 73% were perivenous. At resolutions equivalent to 3T and 1.5T MRI the proportion of perivenous lesions was 67% and 59% respectively. The image resolution did not significantly affect the diagnosis of MS, but the threshold used to discriminate MS from other conditions significantly decreased (p < 0.05) at low resolutions. At a resolution equivalent to 3T MRI, the proportion of perivenous lesions discriminated MS from NMOSD with specificity 87%, sensitivity 93% (55 – 100%), CSVD with specificity 100%, sensitivity 100% (80 – 100%), migraine with specificity 80%, sensitivity 93 (61 – 100%) and inflammatory vasculopathies with specificity 85%, sensitivity 90% (65 – 100%).

Conclusion: The discrimination of MS from its mimics using the CVS should enable rapid translation of this diagnostic marker into routine clinical practice. The threshold used to discriminated MS from its mimics depends on the imaging protocols sensitivity to central veins.

Last updated on 11th September 2020