Lumen and media-adventitia border detection in IVUS images using texture enhanced deformable model

Area of Science:

• Medical Imaging
• Biomedical Engineering
• Computer Vision

Background:

• Accurate delineation of lumen and media-adventitia (MA) borders in intravascular ultrasound (IVUS) images is crucial for vascular assessment and intervention.
• Manual segmentation of these borders is labor-intensive and prone to variability.

Purpose of the Study:

• To develop and validate a novel texture-enhanced deformable model (TEDM) for automated and accurate detection of lumen and MA borders in IVUS images.
• To improve the efficiency and precision of vascular structure measurement and 3D visualization for clinical applications.

Main Methods:

• A texture-enhanced deformable model (TEDM) was developed, integrating texture features with deformable model morphology.
• An ensemble support vector machine classifier was employed to categorize IVUS pixels based on texture features.
• Image regionalization maps derived from tissue classification were used to enhance the TEDM's texture analysis.

Main Results:

• The TEDM method was evaluated on 1500 images from 15 clinical IVUS datasets.
• The automated segmentation achieved high accuracy, with mean surface distance errors of 0.17 mm for lumen and 0.19 mm for MA borders compared to manual delineations.
• The segmentation results enable precise 2D measurements of MA/lumen areas and facilitate 3D vessel reconstructions.

Conclusions:

• The proposed TEDM offers an accurate and efficient automated solution for lumen and MA border segmentation in IVUS imaging.
• This method has the potential to enhance diagnostic capabilities and streamline navigation during vascular interventions.