Research On Key Techniques Of Digital Breast Tomosynthesis Imaging

Breast cancer has become the top cancer in women.Breast screening is the key to prevent and control the breast cancer.Mammography is the first choice for early detection of breast cancer,which has showed high spatial resolution and low radiation dose.However,mammography could not overcome the problem of tissue overlapping.Digital Breast Tomosynthesis(DBT)is an emerging imaging modality based on digital mammography.Limited number of projections are acquired while the X-Ray source is exposured in different orientations at the narrow angular range.The tomographic image can effectively solve the problem of tissue overlapping in mammography and provide more diagnostic data.However there are few researches on DBT systems and imaging reconstruction method investigated in China,the research about 2D/3D imaging techniques of the DBT was carried out in this paper,the main works are as follows:First,a multi-scale contrast enhancement algorithm for mammogram based on LP-DUM is proposed in this paper.The dynamic weights calculated by the local and global mean attenuation density of mammary gland is applied to dynamic unsharp masking(DUM)for de-noising and edge enhancement during the Lapalace Pyramid(LP)decomposition.The enhancement result of phantom and mammography images have showed that the contrast of the proposed method was well improved and no excessive noise was introduced.The contour and edge of the breast were significantly reserved with more detail and high imformation entropy.Second,a fast iteration reconstruction algorithm based on SATpV is proposed.For solving the problems of undersampled data reconstruction and long reconstruction time,an acceleration method for the computation of weights for forward-projection and back-projection are proposed.SART algorithm with GPU acceleration is applied to 3D DBT reconstruction.Meanwhile,a selected anisotropic total p variation is proposed for DBT image regulation.Projection reordering is applied to projection updating for speeding up convergence rate.The experimental results of phantom and prototype data show that the proposed method has the best reconstruction quality with no obvious distortion.The convergence speed is the fastest compared with other methods,and reconstruction time of five iteration reconstruction is less than 9 seconds.The time efficiency and image quality is well guaranteed for clinical acceptance.Third,a geometric calibration method based on image quality evaluation is proposed in this paper.Geometry artifact in reconstructed image is caused by random error of mechanical installation and motor vibration.2-norm of gradiant of image which is reconstructed by the initial geometric parameters solved by the analytic algorithm,is applied to be objective function and optimized for the optimal geometric set in iteration stage.The results of simulation and prototype experiment show that the proposed algorithm can significantly improve the image quanlity in both CBCT and DBT system and the accuracy of geometric parameters effectively,which can also reduce the processing difficulty of the calibration phantom.Fourth,a truncation artifact reduction based on global projection correction and image reconstruction algorithm for DBT is developed.Parts of the projection data would be truncated due to the finite size of detector and the large angles of exposure,which can cause truncation artifact in the reconstructed image.In this paper,the difference of truncation volume of adjacent projections would be calculated,which would be corrected with the reconstructed image by weighted backprojection and the correction term of neighbor region.Then the forward-projection and back-projection weighted update are sequentially preformed during the iteration.The simulated and realistic results have demonstrated that the proposed method can effectively suppress the truncation artifacts and significantly improve the reconstructed image quality.In this paper,we have presented initial achievements in 2D image contrast enhancement,3D image reconstruction,geometric calibration and truncation artifact reduction.However,due to experimental conditions and study time,the relevant research is still in the initial stage and will be further investigated and studied in the next stage.