Automatic Segmentation of Abdominal Wall in Ventral Hernia CT: A Pilot Study

Area of Science:

• Medical Imaging
• Computational Anatomy
• Surgical Technology

Background:

• Ventral hernia (VH) repair has high recurrence rates (24-43%), often due to limitations in current imaging guidance.
• Computed tomography (CT) is used for VH intervention, but relies on qualitative clinical judgment, lacking quantitative metrics.
• There is a need for objective, image-based measurements to characterize abdominal wall anatomy and hernias.

Purpose of the Study:

• To develop and validate automated image segmentation methods for quantifying abdominal wall geometry.
• To establish a foundation for measuring hernia and surrounding tissue properties to optimize VH intervention.
• To introduce a novel approach for geometric classification of the abdominal wall using CT scans.

Main Methods:

• A pilot study utilizing four clinically acquired CT scans from post-operative patients.
• A hierarchical image segmentation approach using level set methods to isolate the abdominal wall.
• Inclusion of bony landmarks and skin surfaces within the segmentation process for contextualization.

Main Results:

• Demonstrated a novel approach for geometric classification of the abdominal wall and key abdominal features.
• Quantitative validation of segmentation results against manually labeled ground truth.
• Achieved a mean surface error of less than 2mm for the outer abdominal wall surface.

Conclusions:

• The developed image segmentation method provides accurate geometric characterization of the abdominal wall.
• This quantitative approach establishes a baseline for improving ventral hernia care and surgical planning.
• Automated segmentation offers a promising tool to overcome limitations of qualitative assessments in VH treatment.