The segmentation and registration of structures are gaining importance due to the increasing demand of automated image enhancement and understanding. In neuro vascular surgery often Indocyanine green (ICG) is used as dye to visualize blood in the vessel. ICG binds to the proteins of the blood plasma, emits in the Infra-Red (IR) spectral band and therefore can be used with the help of an IR- camera to visualize blood within a vessel without any contact (see figure to the left).
This type of angiography is virtually a segmentation. This valuable information in the ICG-IR domain shall be used and transferred to RGB domain (figure to the right) to visualize the vessel intraoperatively to the surgeon. Once registered, other investigations of the vessel can be done. This would be helpful to support the surgeon, his assistants and facilitate the surgical workflow.
The project consists of a literature research on robust registration of structures in a gray scale image, similarities in both domains and a robust registration of the vessel in the RGB domain. The implementation is done in MATLAB.
Robust und (semi-) automated segmentation of the vessel (template generation) in the ICG-IR domain (left figure). Transfer of information from ICG-IR to RGB domain. Automated registration of the vessel in the RGB domain with an adaptive template.