Tongue volume in spinal and bulbar muscular atrophy (SBMA): an AI-assisted automatic MRI analysis.
Authors
Affiliations (5)
Affiliations (5)
- Department of Neurology, University Hospital Ulm, Oberer Eselsberg 45, 89081, Ulm, Germany. [email protected].
- Department of Internal Medicine II, Ulm University Medical Center, Ulm, Germany.
- Department of Neurology, University Hospital Ulm, Oberer Eselsberg 45, 89081, Ulm, Germany.
- German Center for Neurodegenerative Diseases (DZNE), Ulm, Germany.
- Core Facility Small Animal MRI, University of Ulm, Ulm, Germany.
Abstract
Atrophy of the tongue muscle without severe dysarthria is one of the clinical hallmarks of spinal and bulbar muscular atrophy (SBMA), a motor neuron disease caused by an androgene receptor defect. An operator-independent AI-based automatic segmentation of the tongue was applied to 3-D MRI data of the head in SBMA in order to quantify the tongue atrophy. Thirty-nine patients with SBMA and 51 age-matched healthy controls underwent MRI which were used for tongue volume quantification. A single triplanar convolutional neural network of U-Net architecture trained on axial, coronal, and sagittal planes was used for the segmentation of the tongue in MRI scans of the head, the resulting volumes were processed slice-wise across the three orientations and corrected for age. At the group level, a significant atrophy of the tongue was observed in SBMA when compared to controls (p < 0.05). Atrophy correlated well with total SBMA-functional rating scale and even more with bulbar subscores. In summary, the study employed an AI-assisted advanced imaging analysis to quantify the tongue morphology in individuals with SBMA in correlation to clinical bulbar function, suggesting this approach as a potential biomarker for disease assessment.