[A fast adaptive active contour model based on local gray difference for parotid duct].

Area of Science:

• Medical Imaging
• Image Segmentation
• Computational Anatomy

Background:

• Accurate segmentation of the parotid duct is crucial for diagnosing and treating salivary gland diseases.
• Existing active contour models often struggle with complex image gradients and uneven illumination, leading to suboptimal segmentation results.
• The Local Binary Fitting (LBF) model provides a foundation but requires enhancements for adaptive performance.

Purpose of the Study:

• To develop a fast adaptive active contour model for precise parotid duct image segmentation.
• To improve segmentation accuracy and efficiency by incorporating local gray differences.
• To enhance robustness against image artifacts like uneven gray levels and blurred boundaries.

Main Methods:

• Modified the LBF model by introducing a local gray difference as an energy term for the driving evolution curve.
• Utilized local gray-scale variance difference to control energy parameters, replacing traditional λ1 and λ2.
• Incorporated two local similarity factors with varying neighborhood sizes to mitigate effects of uneven image gray levels and boundary blur.

Main Results:

• The proposed algorithm adaptively adjusts evolution direction, velocity, and energy weights based on local gray mean and variance differences.
• Successfully detected target boundaries even in regions with complex gradients.
• Demonstrated rapid and accurate convergence of the evolution curve to the actual parotid duct boundary.

Conclusions:

• The developed adaptive active contour model significantly outperforms existing segmentation algorithms for parotid duct imaging.
• Achieves fast and accurate segmentation while preserving fine image details.
• Offers a superior solution for parotid duct image segmentation tasks.