Related Experiment Video
Updated: May 11, 2026
13:44
Detection of Architectural Distortion in Prior Mammograms via Analysis of Oriented Patterns
Published on: August 30, 2013
Mammographical mass detection and classification using local seed region growing-spherical wavelet transform
Pelin Görgel1, Ahmet Sertbas, Osman N Ucan
1Department of Computer Engineering, Faculty of Engineering, Istanbul University (IU), Istanbul, Turkey. paras@istanbul.edu.tr
Computers in Biology and Medicine
|May 15, 2013
Summary
This study introduces the Local Seed Region Growing-Spherical Wavelet Transform (LSRG-SWT) for accurate breast mass detection and classification in mammograms. The LSRG-SWT method achieved high accuracy in distinguishing benign from malignant masses.
Area of Science:
- Medical Imaging
- Computer-Aided Diagnosis
- Biomedical Engineering
Background:
- Accurate detection and classification of breast masses are crucial for effective diagnosis and treatment.
- Existing mammogram analysis methods require improvement in accuracy and efficiency.
Purpose of the Study:
- To develop and validate a novel computer-aided diagnostic tool for precise detection and classification of breast masses.
- To implement an accurate method for distinguishing between benign and malignant breast masses using mammograms.
Main Methods:
- The proposed Local Seed Region Growing-Spherical Wavelet Transform (LSRG-SWT) method involves four steps: homomorphic filtering, region of interest (ROI) detection using Local Seed Region Growing (LSRG), Spherical Wavelet Transform (SWT) with feature extraction, and Support Vector Machine (SVM) classification.
- The method includes a two-component classification: first, distinguishing ROIs as mass or non-mass, and second, classifying masses as benign or malignant.
Main Results:
- The LSRG-SWT scheme achieved 96% accuracy for mass/non-mass classification and 93.59% for benign/malignant classification on the Istanbul University (I.U.) database using k-fold cross-validation.
- External validation using the I.U. database for training and the Mammographic Image Analysis Society (MIAS) database for testing yielded 94% and 91.67% accuracy for mass/non-mass and benign/malignant classification, respectively.
Conclusions:
- The proposed LSRG-SWT method demonstrates high accuracy and potential as a reliable diagnostic tool for breast mass detection and classification in mammography.
- The study highlights the effectiveness of combining LSRG for ROI detection with SWT and SVM for robust classification of breast lesions.
