Level set method coupled with Energy Image features for brain MR image segmentation.

Area of Science:

• Medical Imaging
• Computer Vision
• Image Processing

Background:

• Brain magnetic resonance (MR) image segmentation faces significant challenges due to noise and intensity inhomogeneity.
• These drawbacks limit the accuracy and reliability of automated segmentation methods.

Purpose of the Study:

• To introduce and evaluate an energy image feature approach for correcting intensity inhomogeneity in brain MR image segmentation.
• To enhance the robustness of segmentation techniques, especially for brain lesion identification.

Main Methods:

• Developed an energy image feature by replacing pixel values with local energy values computed using a 3x3 mask.
• Integrated this energy image feature into a region-based active contours framework utilizing level set methods.
• Compared performance using two level set variants: intensity fitting and selective filtering regularized methods.

Main Results:

• The energy image feature approach demonstrated flexibility in adapting to the level set segmentation framework.
• The method proved effective in robustly segmenting brain lesions, overcoming common segmentation drawbacks.
• Comparative analysis showed the utility of energy image features in improving segmentation accuracy.

Conclusions:

• The energy image feature approach offers a robust solution for intensity inhomogeneity correction in brain MR image segmentation.
• This technique enhances the performance of level set methods, particularly for challenging tasks like brain lesion segmentation.
• The findings suggest a significant advancement in the field of medical image analysis and segmentation.