A Boosted Minimum Cross Entropy Thresholding for Medical Images Segmentation Based on Heterogeneous Mean Filters...

Area of Science:

• Medical Imaging
• Computer Vision
• Image Processing

Background:

• Accurate foreground detection in medical images is crucial for early disease diagnosis.
• Image segmentation methods directly impact diagnostic precision.
• Thresholding techniques, like Minimum Cross Entropy Thresholding (MCET), are common for medical image segmentation.

Purpose of the Study:

• To enhance the efficiency of the Minimum Cross Entropy Thresholding (MCET) method for medical image segmentation.
• To improve foreground detection accuracy by optimizing the MCET algorithm.
• To reduce the impact of noise and outliers in medical image segmentation.

Main Methods:

• Developed a novel approach using heterogeneous mean filter techniques to refine MCET.
• The proposed model estimates an optimized mean by mitigating noise and local outliers.
• Implemented a modified objective function for MCET utilizing the optimized mean values.

Main Results:

• The enhanced MCET method demonstrated improved performance compared to original and related techniques.
• Evaluated using three distinct medical image datasets.
• Achieved accurate segmentation results validated by unsupervised and supervised performance metrics.

Conclusions:

• The proposed heterogeneous mean filter-based MCET significantly boosts segmentation accuracy in medical images.
• This advancement contributes to more precise feature extraction and improved diagnostic capabilities.
• The optimized method offers a robust solution for foreground-background separation in medical imaging applications.