Related Experiment Video
Updated: May 17, 2026
13:44
Detection of Architectural Distortion in Prior Mammograms via Analysis of Oriented Patterns
Published on: August 30, 2013
Multiresolution local binary pattern texture analysis combined with variable selection for application to
1Image and Video Systems Lab, Department of Electrical Engineering, Korea Advanced Institute of Science and Technology-KAIST 335, Gwahak-ro, Yuseon-gu, Daejeon 305-701, Korea. Jae.YoungChoi@uphs.upenn.edu
Physics in Medicine and Biology
|October 12, 2012
Summary
A novel multiresolution Local Binary Pattern (LBP) feature effectively characterizes mass textures to reduce false-positive (FP) detection in mammography. Combining LBP features with variable selection enhances computer-aided detection (CAD) of breast masses.
Area of Science:
- Medical Imaging
- Computer Vision
- Biomedical Engineering
Background:
- Computer-aided detection (CAD) systems aim to reduce false-positive (FP) findings in mammography.
- Accurate characterization of mass texture is crucial for differentiating malignant from benign lesions.
Purpose of the Study:
- To develop a novel texture feature for improved mass characterization in mammography.
- To reduce false-positive detection rates in computer-aided mass detection frameworks.
Main Methods:
- Extraction of multiresolution Local Binary Pattern (LBP) texture features from segmented mass regions.
- Incorporation of multiresolution texture analysis to preserve spatial information.
- Application of Support Vector Machine-Recursive Feature Elimination (SVM-RFE) for optimal feature selection.
Main Results:
- The proposed multiresolution LBP features effectively characterize mass textures, including core and margin regions.
- Experimental results on benchmark mammogram databases demonstrate superior performance compared to existing texture features for FP reduction.
- Combining multiresolution LBP features with variable selection significantly reduces false-positive signals.
Conclusions:
- The novel multiresolution LBP feature is a promising tool for enhancing the accuracy of computer-aided detection of mammographic masses.
- The proposed method offers an effective solution for reducing false-positive findings, improving the reliability of CAD systems.
