基于深度学习的纵向MRI胶质瘤分组配准
图像与视频处理
2023-06-21 v1 计算机视觉与模式识别
机器学习
摘要
胶质瘤的生长可通过纵向图像配准进行量化。然而，图像间较大的占位效应与组织变化带来了额外的挑战。在此，我们提出一种纵向的、基于学习的、分组式的配准方法，用于胶质瘤MRI的准确且无偏的配准。我们在来自Glioma Longitudinal AnalySiS联盟的数据集上进行评估，并将其与经典配准方法进行比较。我们取得了相当的Dice系数，且配准结果更为精细，同时将运行时间显著缩短至一分钟内。所提出的方法可作为经典工具箱的替代方案，以进一步揭示胶质瘤的生长规律。
引用
@article{arxiv.2306.10611,
title = {Deep learning-based group-wise registration for longitudinal MRI analysis in glioma},
author = {Claudia Chinea Hammecher and Karin van Garderen and Marion Smits and Pieter Wesseling and Bart Westerman and Pim French and Mathilde Kouwenhoven and Roel Verhaak and Frans Vos and Esther Bron and Bo Li},
journal= {arXiv preprint arXiv:2306.10611},
year = {2023}
}
备注
Digital poster presented at the annual meeting of the International Society for Magnetic Resonance in Medicine (ISMRM) 2023. A 6 minute video about this work is available for browsing by the conference website (Program number: 4361)