Improved mass detection in mammogram images with Dual Tree Complex Wavelet Transform and Fourier Descriptors
M. Kanchana, Ramu Naresh, C. N. S. Vinoth Kumar, P. Pandiaraja · Artificial Intelligence in Medicine · 2024
Breast cancer is ranked second when compared with other cancers. Mammography is the widely accepted technique, which was recognized as a gold standard for the detection of breast tumor disease. In this work, we are going to identify the mass in mammogram images using Dual Tree Complex Wavelet Transform (DTCWT), which is used for image decomposition; feature extraction is done with the help of Fourier Descriptor and Artificial Neural Network (ANN) used for classification tasks. Among the several databases, the most frequently used databases are the MIAS (Mammographic Image Analysis Society) database and DDSM (Digital Database for Screening Mammography). In our work, we utilized the MIAS database and its mammogram image description for evaluation purposes. Dual-tree complex discrete wavelet transform benefits are low computation time, limited redundancy, perfect reconstruction, good shift invariance, and directional selectivity. As a result, the proposed work achieves classification accuracy of 92.3%, sensitivity of 96.3%, and specificity of 94.5%.