Multi-feature medical image registration for image-guided neurosurgery

在图像引导的神经外科手术中，目前存在许多配准技术，可以将来自不同成像方式的互补信息（即解剖、功能、化学或血管）与标准解剖图谱相结合，进行详细分析，以便计划和实施手术干预。然而，这些技术在存在病理或切除的情况下不起作用。我们的目标是开发和测试新的非线性n维配准方法，使不连续变形成为可能。这些方法将改善手术指导，使干预更准确、侵入性更小，降低患者发病率、住院时间和费用。

In image-guided neurosurgery, many registration techniques currently exist to enable complementary information (i.e., anatomical, functional, chemical or vascular) from different imaging modalities to be combined with standard anatomical atlases for detailed analysis in order to plan and carry out the surgical intervention.  However, these techniques do no work well in the presence of pathology or resections. Our goal is to develop and test new non-linear N-dimensional registration methods that are enabled for discontinuous deformations.  Such methods will improve surgical guidance and result in interventions that are more accurate and less invasive, and patient morbidity, hospitalization time and cost will be reduced.