Conventional mammographic image generation in dual-energy digital mammography

摘要

Dual-energy digital mammography (DEDM) can generate tissue-subtracted calcification image for improving the detectability of breast calcifications. However, the masses, if present, are missing in the tissue-subtracted calcification image. This paper proposes an algorithm to generate conventional mammographic image by DEDM images based on a multi-scale decomposition and reconstruction architecture with Gaussian filters. Firstly, calibration coefficients are measured at different kVp to correct the original LE and HE. Secondly, the LE and HE images are decomposed into multi-scale components. Thirdly, the components at different scale of the two images are weighted based on a similarity measure to generate new components. Finally, the conventional mammographic image is reconstructed by these new components using noise suppression technique. The proposed method was validated by different breast phantoms on two commercially available full-field digital mammography systems. Results show that the method is effective and the reconstructed image has similar grayscale, contrast and noise level to the corresponding conventional mammogram. Therefore, both the calcification image and conventional mammogram-like image can be generated; the patient will not need more exposure to get the conventional mammogram in DEDM.