A robust medical image segmentation method using KL distance and local neighborhood information

Area of Science:

• Medical image analysis
• Computer vision
• Image segmentation

Background:

• Level set segmentation models like Chan-Vese (CV) struggle with image heterogeneity and complex structures.
• Accurate medical image segmentation is often hindered by similar intensity regions and requires manual parameter tuning.
• Existing CV models may not achieve optimal results in medical imaging without significant user intervention.

Purpose of the Study:

• To propose an improved Chan-Vese (CV) model for more accurate and robust image segmentation.
• To address limitations of traditional CV models in handling image heterogeneity and complex object boundaries.
• To enhance medical image segmentation by incorporating local image characteristics.

Main Methods:

• Developed an improved Chan-Vese (CV) model integrating Kullback-Leibler (KL) distances.
• Incorporated local neighborhood information (LNI) to capture image-specific characteristics.
• Validated the method using synthetic and real medical image datasets.

Main Results:

• The proposed method demonstrated superior segmentation accuracy compared to state-of-the-art techniques.
• Experimental results showed enhanced robustness in segmenting images with complex structures and heterogeneity.
• The improved model achieved higher efficiency in medical image segmentation tasks.

Conclusions:

• The enhanced CV model effectively overcomes limitations of traditional methods in medical image segmentation.
• Integration of KL distances and LNI significantly improves segmentation accuracy and robustness.
• The proposed approach offers a more efficient and reliable solution for medical image segmentation.