Improving segmentation procedure

Trying to improve the segmentation of carotid arteries with time-of-flight (TOF) MR images, and I sometimes get a lot of artefact within the segmentation. This looks to be fatty tissue within the neck. I am using these images to calculate tortuosity. Sometimes, the artefact doesn’t affect the centerline extract that I perform afterwards, however sometimes it does. How can I improve the segmentation procedure so I can make the tortuosity measurements more accurate?

Additionally, as I am creating a segmentation of the extracranial carotid (common, external and internal carotid), sometimes the centerline calculates using the ECA, when I specifically set one of the seeds at the ICA. How can I improve this?