Choroid Plexus Segmentation Using Optimized 3D U-Net

• Published in: Proceedings (International Symposium on Biomedical Imaging) pp. 381 - 384
• Main Authors: Zhao, Li, Feng, Xue, Meyer, Craig H., Alsop, David C.
• Format: Conference Proceeding
• Language: English
• Published: IEEE 01.04.2020
• Subjects: Choroid Plexus; MRI; Segmentation; U-Net
• ISSN: 1945-8452
• DOI: 10.1109/ISBI45749.2020.9098443

The choroid plexus is the primary organ that secretes the cerebrospinal fluid. Its structure and function may be associated with the brain drainage pathway and the clearance of amyloid-beta in Alzheimer's Disease. However, choroid plexus segmentation methods have rarely been studied. Therefore, the purpose of this work is to fill the gap using a deep convolutional network. MR images of 10 healthy subjects ( 75.5\pm 8.0 years) were retrospectively selected from the Alzheimer's Disease Neuroimaging Initiative database (ADNI). The benchmark of choroid plexus segmentation was provided by the FreeSurfer package and manual correction. A 3D U-Net was developed and optimized in the patch extraction, augmentation, and loss function. In leave-one-out cross-validations, the optimized U-Net provided superior performance compared to the FreeSurfer results (Dice score 0.732\pm 0.046 vs 0.581\pm 0.093 , Jaccard coefficient 0.579\pm 0.057 vs 0.416\pm 0.091 , 95% Hausdorff distance 1.871\pm 0.549 vs 7.257\pm 5.038 , and sensitivity 0.761 \pm 0.078 vs 0.539\pm 0.117 ).