Banerjee Rohan, Kaptan Merve, Tinnermann Alexandra, Khatibi Ali, Dabbagh Alice, Büchel Christian, Kündig Christian W, Law Christine S W, Pfyffer Dario, Lythgoe David J, Tsivaka Dimitra, Van De Ville Dimitri, Eippert Falk, Muhammad Fauziyya, Glover Gary H, David Gergely, Haynes Grace, Haaker Jan, Brooks Jonathan C W, Finsterbusch Jürgen, Martucci Katherine T, Hemmerling Kimberly J, Mobarak-Abadi Mahdi, Hoggarth Mark A, Howard Matthew A, Bright Molly G, Kinany Nawal, Kowalczyk Olivia S, Freund Patrick, Barry Robert L, Mackey Sean, Vahdat Shahabeddin, Schading Simon, McMahon Stephen B, Parish Todd, Marchand-Pauvert Véronique, Chen Yufen, Smith Zachary A, Weber Ii Kenneth A, De Leener Benjamin, Cohen-Adad Julien
Department of Computer Science, Polytechnique Montreal, Montreal, Quebec, Canada.
Mila-Quebec AI Institute, Montreal, Quebec, Canada.
Imaging Neurosci (Camb). 2025 Sep 9;3. doi: 10.1162/IMAG.a.98. eCollection 2025.
Functional magnetic resonance imaging (fMRI) of the spinal cord is relevant for studying sensation, movement, and autonomic function. Preprocessing of spinal cord fMRI data involves segmentation of the spinal cord on gradient-echo echo planar imaging (EPI) images. Current automated segmentation methods do not work well on these data, due to the low spatial resolution, susceptibility artifacts causing distortions and signal drop-out, ghosting, and motion-related artifacts. Consequently, this segmentation task demands a considerable amount of manual effort which takes time and is prone to user bias. In this work, we (i) gathered a multi-center dataset of spinal cord gradient-echo EPI with ground-truth segmentations and shared it on OpenNeuro https://openneuro.org/datasets/ds005143/versions/1.3.1 and (ii) developed a deep learning-based model, EPISeg, for the automatic segmentation of the spinal cord on gradient-echo EPI data. We observe a significant improvement in terms of segmentation quality compared with other available spinal cord segmentation models. Our model is resilient to different acquisition protocols as well as commonly observed artifacts in fMRI data. The training code is available at https://github.com/sct-pipeline/fmri-segmentation/, and the model has been integrated into the Spinal Cord Toolbox as a command-line tool.
脊髓功能磁共振成像（fMRI）对于研究感觉、运动和自主神经功能具有重要意义。脊髓fMRI数据的预处理涉及在梯度回波平面回波成像（EPI）图像上对脊髓进行分割。由于空间分辨率低、易感性伪影导致图像失真和信号丢失、鬼影以及与运动相关的伪影，当前的自动分割方法在这些数据上效果不佳。因此，这种分割任务需要大量的人工操作，既耗时又容易出现用户偏差。在这项工作中，我们（i）收集了一个带有真实分割的脊髓梯度回波EPI多中心数据集，并在OpenNeuro（https://openneuro.org/datasets/ds005143/versions/1.3.1）上共享，（ii）开发了一种基于深度学习的模型EPISeg，用于在梯度回波EPI数据上自动分割脊髓。与其他可用的脊髓分割模型相比，我们观察到在分割质量方面有显著提高。我们的模型对不同的采集协议以及fMRI数据中常见的伪影具有弹性。训练代码可在https://github.com/sct-pipeline/fmri-segmentation/获取，并且该模型已作为命令行工具集成到脊髓工具箱中。
