Automatic vessel lumen segmentation and stent strut detection in intravascular optical coherence tomography

Area of Science:

• Medical Imaging
• Biomedical Engineering
• Cardiovascular Research

Background:

• Optical coherence tomography (OCT) provides high-resolution intravascular imaging.
• Accurate segmentation of lumen area and stent struts is crucial for quantitative analysis of neointimal hyperplasia (NIH).
• Manual analysis of OCT images is time-consuming and prone to variability.

Purpose of the Study:

• To develop an automated segmentation technique for lumen area extraction and stent strut detection in intravascular OCT images.
• To enable quantitative morphological analysis of in-stent neointimal hyperplasia (NIH).
• To improve the efficiency and accuracy of OCT image analysis.

Main Methods:

• A Markov random field (MRF) model was used for lumen area segmentation.
• Textural and edge information from local intensity distribution and continuous wavelet transform (CWT) were integrated for contour extraction.
• Stent strut positions were detected using a feature extraction and classification scheme based on wavelet responses.

Main Results:

• The automated segmentation achieved high accuracy in extracting the inner lumen contour and stent strut positions.
• The average overlap value for lumen segmentation compared to manual analysis was 0.937 ± 0.045.
• Stent strut detection demonstrated high performance with an AUC of 0.95, sensitivity of 0.91, and specificity of 0.96.

Conclusions:

• A robust automatic segmentation technique for intravascular OCT images was developed.
• The algorithm effectively integrates textural and edge information for vessel lumen border and stent strut detection.
• This automated approach facilitates quantitative morphological analysis of in-stent neointimal hyperplasia.