AUTHOR=Zhang Chuan , Li Man , Luo Zheng , Xiao Ruhui , Li Bing , Shi Jing , Zeng Chen , Sun BaiJinTao , Xu Xiaoxue , Yang Hanfeng TITLE=Deep learning-driven MRI trigeminal nerve segmentation with SEVB-net JOURNAL=Frontiers in Neuroscience VOLUME=Volume 17 - 2023 YEAR=2023 URL=https://www.frontiersin.org/journals/neuroscience/articles/10.3389/fnins.2023.1265032 DOI=10.3389/fnins.2023.1265032 ISSN=1662-453X ABSTRACT=Purpose: Trigeminal neuralgia (TN) poses significant challenges in its diagnosis and treatment due to its extreme pain. Magnetic Resonance Image (MRI) plays a crucial role in diagnosing TN and understanding its pathogenesis. Manual delineation of the trigeminal nerve in volumetric images is time-consuming and subjective. This study introduces a Squeeze and Excitation with BottleNeck V-Net (SEVB-Net), a novel approach for automatic segmentation of the trigeminal nerve in threedimensional T2 MRI volumes.We enrolled 88 patients with trigeminal neuralgia and 99 healthy volunteers, dividing them into training and testing groups. The SEVB-Net was designed for end-to-end training, taking three-dimensional T2 images as input and producing segmention volume of the same size. We assessed the performance of the basic V-Net, nnUNet and SEVB-Net model by calculating the dice similarity coefficient (DSC), sensitivity, precision and network complexity. Additionally, we used the Mann-Whitney U test to compare the time required for manual segmentation and automatic segmentation with manual manual modification.In the testing group, the experimental results demonstrated that the proposed method achieved state-of-the-art performance. SEVB-Net combined with ω DoubleLoss loss function, achieved a DSC ranging from 0.6070 to 0.7923. SEVB-Net combined with ωDoubleLoss method and nnUNet combined with DoubleLoss method, respectively, achieved DSC, sensitivity, and precision values exceeding 0.7. However, SEVB-Net significantly reduced the number of parameters (2.20M), memory consumption (11.41MB), and model size (17.02MB), resulting in improved computation and forward time compared to nnUNet. The difference in average time between manual segmentation and automatic segmentation with manual modification for both radiologists was statistically significant (p<0.001).The experimental results demonstrate that the proposed method can automatically segment the root and three main branches of the trigeminal nerve in three-dimensional T2 images. SEVB-Net, compared to the basic V-Net model, showed improved segmentation performance and achieved a level similar to nnUNet. Both SEVB-Net and nnUNet's segmentation volumes aligned with expert annotations, but SEVB-Net displayed a more lightweight feature.