Temporal Dynamics-Driven Automatic Segmentation of Parasagittal Dural Space and Meningeal Lymphatic Vessels in DCE-MRI Using VQ-VAE

Motivation: Manual delineation of meningeal lymphatic vessels (mLVs) in the parasagittal dural space (PSD) from adjacent structures in DCE-MRI is challenging due to low spatial resolution and partial volume effects. Goal(s): Our goal was to develop an automated method for segmenting the PSD and mLVs using the temporal dynamics of DCE-MRI. Approach: We used continuous wavelet transform and employed a vector-quantized variational autoencoder (VQ-VAE) for the automatic segmentation of the PSD and mLVs using the temporal dynamics of DCE-MRI. Results: Semi-quantitative parameters obtained from mLVs differed from those of adjacent structures, which may be used for more accurate assessment of glymphatic clearance capacity. Impact: Our proposed method may enable automatic segmentation of the PSD and mLVs in cases of highly limited spatial resolution, where manual segmentation may be impractical.