Background Parkinsonian variant of multiple system atrophy is a neurodegenerative disorder frequently misdiagnosed as Parkinson’s disease. No early imaging biomarkers currently differentiate these disorders. Methods Simple visual imaging analysis of the substantia nigra and locus coeruleus in neuromelanin-sensitive magnetic resonance imaging and nigrosome 1 in susceptibility-weighted sequences was performed in thirty patients with parkinsonian variant of multiple system atrophy fulfilling possible/probable second consensus diagnostic criteria. The neuromelanin visual pattern was compared to patients with Parkinson’s disease with the same disease duration (n = 10) and healthy controls (n = 10). Substantia nigra semi-automated neuromelanin area/signal intensity was compared to the visual data. Results Groups were similar in age, sex, disease duration, and levodopa equivalent dose. Hoehn & Yahr stage was higher in parkinsonian multiple system atrophy patients, 69% of whom had normal neuromelanin size/signal, significantly different from Parkinson’s disease patients, and similar to controls. Nigrosome 1 signal was lost in 74% of parkinsonian multiple system atrophy patients. Semi-automated neuromelanin substantia nigra signal, but not area, measurements were able to differentiate groups. Conclusions In patients with parkinsonism, simple visual magnetic resonance imaging analysis showing normal neuromelanin substantia nigra and locus coeruleus, combined with nigrosome 1 loss, allowed the distinction of the parkinsonian variant of multiple system atrophy from Parkinson’s disease and healthy controls. This easy and widely available method was superior to semi-automated measurements in identifying specific imaging changes in substantia nigra and locus coeruleus.

We present a deep learning method for the segmentation of new lesions in longitudinal FLAIR MRI sequences acquired at two different time points. In our approach, the 3D volumes are processed slice-wise across the coronal, axial, and sagittal planes and the predictions from the three orientations are merged using an optimized voting strategy. Our method achieved best F1 score (0.541) among all participating methods in the MICCAI 2021 challenge Multiple sclerosis new lesions segmentation (MSSEG-2). Moreover, we show that our method is on par with the challenge's expert neuroradiologists: on an unbiased ground truth, our method achieves results comparable to those of the four experts in terms of detection (F1 score) and segmentation accuracy (Dice score).

No abstract