Shape prior generation and geodesic active contour interactive iterating algorithm (SPACIAL): fully automatic...

Area of Science:

• Medical Imaging
• Image Segmentation
• Cardiovascular Imaging

Background:

• Intravascular optical coherence tomography (OCT) enables detailed vessel imaging.
• Accurate lumen segmentation is crucial for assessing cardiovascular health.
• Image quality in OCT is often compromised by artifacts, hindering segmentation.

Purpose of the Study:

• To develop a fully automatic lumen segmentation algorithm for low-quality OCT images.
• To introduce a novel method, SPACIAL, guided by an adaptively generated shape prior.
• To enhance the accuracy and efficiency of vessel lumen segmentation.

Main Methods:

• SPACIAL utilizes an interactive framework where active contours and shape priors iteratively refine segmentation.
• The shape prior is adaptively generated based on the evolving active contour.
• A fast algorithm accelerates 3D image segmentation.

Main Results:

• SPACIAL achieved a high average Dice coefficient of 93.6(2.4)% on 3240 images.
• The algorithm demonstrated significant time efficiency, being 5.7 times faster than classical level set methods.
• Satisfactory segmentation accuracy and speed were confirmed in experimental validation.

Conclusions:

• SPACIAL provides accurate and efficient lumen segmentation for low-quality OCT images.
• The method holds significant importance for cardiovascular disease diagnosis.
• SPACIAL shows considerable potential for clinical applications in vascular imaging.