Multi-scale Blood Vessel Detection and Segmentation in Breast MRIs

Gilad Kahala*, Miri Sklair, Hedva Spitzer

*Corresponding author for this work

Research output: Contribution to journalArticlepeer-review

Abstract

An algorithm is proposed to perform segmentation of blood vessels in 3D breast MRIs. The blood vessels play an essential role as an additional tool to detect tumors. Large concentration of blood vessels can indicate a malignant mass. Radiologists use a maximum-intensity projection to expose the vasculature. However, the breast is a challenging organ in identifying vascular structures, because of noise and fat tissues. There are several existing algorithms to detect blood vessels in MRI, but those usually prove insufficient when it comes to the breast. Our algorithm provides a three-dimensional model of blood vessels by utilizing texture enhancement, Hessian-based methods and blood vessel completion by center line tracking. We compared the algorithm results to manually segmented images done by radiologists in 24 different patients. They yielded in 86% sensitivity to the ground truth and 88.3% specificity. It also appears that by employing mass detection as the last step, our algorithm can provide a helpful tool for tumor enhancement and automated detection of breast cancer.

Original languageEnglish
Pages (from-to)424-430
Number of pages7
JournalJournal of Medical and Biological Engineering
Volume39
Issue number3
DOIs
StatePublished - 13 Jun 2019
Externally publishedYes

Keywords

  • Blood vessels
  • Breast
  • Center line tracking
  • Magnetic resonance imaging (MRI)
  • Vessel enhancement

Fingerprint

Dive into the research topics of 'Multi-scale Blood Vessel Detection and Segmentation in Breast MRIs'. Together they form a unique fingerprint.

Cite this