An automatic algorithm for detecting stent endothelialization from volumetric optical coherence tomography datasets

Garret T. Bonnema, Kristen O.Halloran Cardinal, Stuart K. Williams, Jennifer K. Barton

Research output: Contribution to journalArticlepeer-review

38 Scopus citations

Abstract

Recent research has suggested that endothelialization of vascular stents is crucial to reducing the risk of late stent thrombosis. With a resolution of approximately 10 μm, optical coherence tomography (OCT) may be an appropriate imaging modality for visualizing the vascular response to a stent and measuring the percentage of struts covered with an anti-thrombogenic cellular lining. We developed an image analysis program to locate covered and uncovered stent struts in OCT images of tissue-engineered blood vessels. The struts were found by exploiting the highly reflective and shadowing characteristics of the metallic stent material. Coverage was evaluated by comparing the luminal surface with the depth of the strut reflection. Strut coverage calculations were compared to manual assessment of OCT images and epi-fluorescence analysis of the stented grafts. Based on the manual assessment, the strut identification algorithm operated with a sensitivity of 93% and a specificity of 99%. The strut coverage algorithm was 81% sensitive and 96% specific. The present study indicates that the program can automatically determine percent cellular coverage from volumetric OCT datasets of blood vessel mimics. The program could potentially be extended to assessments of stent endothelialization in native stented arteries.

Original languageEnglish (US)
Pages (from-to)3083-3098
Number of pages16
JournalPhysics in medicine and biology
Volume53
Issue number12
DOIs
StatePublished - Jun 21 2008

ASJC Scopus subject areas

  • Radiological and Ultrasound Technology
  • Radiology Nuclear Medicine and imaging

Fingerprint Dive into the research topics of 'An automatic algorithm for detecting stent endothelialization from volumetric optical coherence tomography datasets'. Together they form a unique fingerprint.

Cite this