Related Experiment Video
Updated: Mar 1, 2026
13:44
Detection of Architectural Distortion in Prior Mammograms via Analysis of Oriented Patterns
Published on: August 30, 2013
43.7K
Globally supported radial basis function based collocation method for evolution of level set in mass segmentation
Kanchan Lata Kashyap1, Manish Kumar Bajpai1, Pritee Khanna1
1Computer Science & Engineering, PDPM IIITDM, Jabalpur, 482005, India.
Computers in Biology and Medicine
|May 27, 2017
Summary
This study introduces a novel mesh-free level set method for accurate breast mass detection in mammograms. The approach achieves high sensitivity and specificity, outperforming traditional mesh-based methods for improved computer-aided detection.
Area of Science:
- Medical Imaging
- Computer Vision
- Biomedical Engineering
Background:
- Mammography is crucial for breast cancer detection, but mass segmentation is challenging due to image quality and mass characteristics.
- Existing computer-aided detection (CAD) systems face difficulties in accurately segmenting masses with low contrast, irregular shapes, and speculated margins.
Purpose of the Study:
- To develop and evaluate a novel mesh-free radial basis function (RBF) level set approach for enhanced breast mass segmentation in mammograms.
- To compare the performance of the mesh-free method against traditional mesh-based finite difference methods (FDM).
- To assess the efficacy of Binarized Statistical Image Features (BSIF) and Local Binary Patterns (LBP) combined with Support Vector Machine (SVM) for mass classification.
Main Methods:
- Anisotropic diffusion filtering was used for mammogram enhancement.
- A mesh-free RBF collocation method evolved the level set function for segmenting breast and suspicious mass regions.
- Statistical measures, BSIF, and LBP features were computed from segmented regions.
- A Support Vector Machine (SVM) classifier was employed for mass detection.
Main Results:
- The mesh-free RBF approach demonstrated effective segmentation of breast and suspicious mass regions.
- Combined BSIF features yielded superior classification performance compared to LBP variants.
- On the DDSM dataset, the system achieved 97.12% sensitivity, 92.43% specificity, and 98% AUC.
- On the MIAS dataset, the system achieved 95.12% sensitivity, 92.41% specificity, and 95% AUC.
Conclusions:
- The proposed mesh-free level set method offers a robust and accurate solution for breast mass detection in mammography.
- The combination of BSIF features and SVM provides a high-performance classifier for distinguishing between mass and non-mass regions.
- This approach holds significant potential for improving computer-aided detection systems in mammography.
