Abstract #4450

Automated segmentation of human axon and myelin from electron microscopy data using deep learning for microstructural validation and simulation

Qiyuan Tian^1,2, Chanon Ngamsombat¹, Hong-Hsi Lee^3,4, Daniel R. Berger⁵, Yuelong Wu⁵, Qiuyun Fan^1,2, Berkin Bilgic^1,2, Dmitry S. Novikov^3,4, Els Fieremans^3,4, Bruce R. Rosen^1,2, Jeff W. Lichtman⁵, and Susie Y. Huang^1,2

¹Martinos Center for Biomedical Imaging, Massachusetts General Hospital, Charlestown, MA, United States, ²Department of Radiology, Harvard Medical School, Boston, MA, United States, ³Department of Radiology, Center for Biomedical Imaging, New York University School of Medicine, New York, NY, United States, ⁴Center for Advanced Imaging Innovation and Research (CAI2R), New York University School of Medicine, New York, NY, United States, ⁵Department of Molecular and Cellular Biology, Harvard University, Cambridge, MA, United States

Diffusion microstructural metrics represent inferences of axonal size and morphology rather than directly imaged quantities, validation of these metrics is essential. With novelty of multibeam-serial electron microscopy, high-resolution images of human white matter can be acquired at nanometer resolution over volumes of tissue large enough to capture the diffusion-MRI dynamics extending over length scales comparable to MRI voxel size. This work presents automated segmentation of serial EM of a sub-volume of human white matter using a 3D convolutional neural network studying variations in axonal diameter over the longest axons within the volume of tissue.

This abstract and the presentation materials are available to members only; a login is required.

Join Here