Improvement of material decomposition accuracy using denoising and deblurring techniques in spectral mammography

With an increase of breast cancer patients, dual-energy mammographic techniques have been advanced for improving diagnostic accuracy. In general, conventional dual-energy techniques increase radiation dose because the techniques are based on double exposures. Dual-energy techniques with photon-counting detectors (PCDs) can be implemented by using a single exposure. However, the images obtained from the dual-energy techniques with the PCDs suffer from statistical noise because the dual-energy measurements were performed with a single exposure, causing a lack of the number of effective photons. Thus, the material decomposition accuracy is decreased, and the image quality is distorted. In this study, denoising and deblurring techniques were iteratively applied to a dual-energy mammographic technique based on a PCD, and we evaluated RMSE, noise, and CNR for the quantitative analysis of material decomposition. The results showed that the RMSE value was about 0.23 times lower for the decomposed images with the denoising and deblurring techniques than that without the denoising and deblurring techniques. The noise and CNR of the decomposed images were averagely decreased and increased by factors of 0.23 and 4.17, respectively, through the denoising and deblurring techniques. But, the iterative application of the debelurring technique slightly increased the RMSE and noise. Therefore, it is considered that the material decomposition accuracy and image quality can be improved by applying the denoising and deblurring techniques with the appropriate iterations.