Related Experiment Videos
Active contours driven by local and global fitted image models for image segmentation robust to intensity
Farhan Akram1, Miguel Angel Garcia2, Domenec Puig1
1Department of Computer Engineering and Mathematics, Rovira i Virgili University, Tarragona, Spain.
Plos One
|April 5, 2017
Summary
This study introduces an advanced region-based active contour method for segmenting intensity-inhomogeneous images, improving accuracy in medical image analysis. The technique effectively handles bias fields in brain MRI scans.
Area of Science:
- Medical Image Analysis
- Computational Imaging
- Biomedical Engineering
Background:
- Intensity inhomogeneity and bias fields in medical images complicate accurate segmentation.
- Existing active contour methods struggle with significant intensity variations.
- Precise segmentation is crucial for quantitative analysis of brain magnetic resonance (MR) images.
Purpose of the Study:
- To develop a robust region-based active contour method for segmenting intensity-inhomogeneous images.
- To address the challenges posed by bias fields in medical image segmentation.
- To improve the accuracy and practical applicability of image segmentation techniques for brain MR imaging.
Main Methods:
- A region-based active contour model utilizing an energy functional with local and global fitted images.
- Introduction of local and global signed pressure force functions to stabilize gradient descent.
- Bias field correction using Gaussian distribution approximation and image division.
- Development and extension of a two-phase to a four-phase model for segmentation.
- Application of a Gaussian kernel for contour regularization, avoiding re-initialization.
Main Results:
- The local fitted term effectively extracts regions with intensity inhomogeneity.
- The global fitted term successfully targets homogeneous regions.
- The proposed method demonstrates superior performance in segmenting synthetic and real brain MR images compared to state-of-the-art techniques.
- Quantitative and qualitative comparisons validate the practical advantages of the segmentation technique.
Conclusions:
- The developed region-based active contour method offers enhanced accuracy for segmenting intensity-inhomogeneous images, particularly brain MR images.
- The integration of local and global fitting terms, along with bias field correction, significantly improves segmentation robustness.
- This technique provides a valuable tool for medical image analysis, offering practical benefits in clinical applications.