Medical image segmentation based on level set and isoperimetric constraint.

Area of Science:

• Medical Image Analysis
• Computational Geometry
• Differential Geometry

Background:

• Level set methods are widely used for image segmentation.
• Incorporating shape constraints into energy functionals aids contour shape control.
• Existing methods may struggle with objects having incomplete or blurred boundaries.

Purpose of the Study:

• To develop a novel level set segmentation method incorporating an isoperimetric shape constraint.
• To improve the accuracy and robustness of image segmentation, particularly for medical applications.
• To ensure object compactness and handle incomplete or blurred boundaries.

Main Methods:

• Integration of an isoperimetric constraint into a level set framework.
• Penalizing the ratio of squared perimeter to enclosed area of active contours.
• Utilizing scale-invariant and shape-expression-free constraints.

Main Results:

• The proposed method ensures object compactness and completes missing/blurred boundary parts.
• It handles objects of various scales without explicit shape parameter estimation.
• Efficient segmentation of lesions in ultrasound, CT, and MR images was achieved.
• Quantitative evaluation demonstrated superior accuracy compared to existing level set methods.

Conclusions:

• The novel isoperimetric constraint level set method offers accurate and robust image segmentation.
• It shows significant potential for lesion segmentation in medical diagnoses and surgical planning.
• The method's ability to handle scale variations and boundary imperfections makes it broadly applicable.