A Minimal Path Searching Approach for Active Shape Model (ASM)-based Segmentation of the Lung

Area of Science:

• Medical Imaging
• Computer-Aided Diagnosis
• Radiology

Background:

• Accurate lung boundary segmentation is crucial for diagnosing respiratory diseases from digital radiographs.
• Conventional Active Shape Models (ASM) face challenges in precise boundary detection due to limitations in searching strategies.

Purpose of the Study:

• To develop and evaluate an improved Active Shape Model (ASM) segmentation method using a minimal path (MP) searching strategy for enhanced lung boundary detection on digital radiographs.

Main Methods:

• Implemented a novel fan-shaped search region and a minimal path (MP) deformable model to drive the segmentation process.
• Incorporated a statistical shape prior and a smooth constraint into the deformable model to maintain shape integrity.
• Quantitatively assessed the ASM-MP method by comparing its segmentation results against manual segmentation on 72 digital radiographs.

Main Results:

• The proposed ASM-MP method achieved a mean distance error of 1.75 ± 0.33 pixels.
• Conventional ASM resulted in a higher mean distance error of 1.99 ± 0.45 pixels.
• The ASM-MP method demonstrated superior accuracy in segmenting lung boundaries.

Conclusions:

• The minimal path (MP) searching strategy significantly improves the accuracy of Active Shape Model (ASM)-based lung segmentation on digital radiographs.
• The ASM-MP method provides a robust and accurate tool for automated lung boundary detection, aiding in radiological assessments.