Projection Data Optimization And Correction Of Digital Breast Tomosynthesis Imaging

Digital Breast Tomosynthesis(DBT)imaging is a new tomography imaging technology based on the traditional tomography geometrical principle.DBT imaging mainly depends on the characteristics of data acquired by flat panel detectors and dynamic imaging.Similar to conventional digital mammography,a compression plate is utilized to reduce breast tissue structure overlap by pressuring the tissue evenly on detectors.During the data acquisition,the X-ray source scan a limited angle around the breast tissue with static detectors and retrospectively reconstruct images of the arbitrary fault plane parallel to the detector.Compared with the conventional digital mammography technique,DBT imaging considers the three-dimensional characteristic of the breast tissue structure and effectively reduce false-positive caused by the overlap of the gland structure.Moreover,DBT imaging technology improves the resolution,especially on the tissue edge and micro-calcification region.Furthermore,compared with full-view scan of Computed Tomography(CT),the novel limited angle scan of DBT imaging can reduce the total radiation dose with higher contrast resolution.In general,the novel technology not only overcomes the overlap of tissue structure in traditional digital mammography imaging but also reduce the radiation dose on human body.DBT imaging technology in the screening and diagnosis of breast cancer have the great prospect.Due to the geometric properties of DBT imaging and the structure particularity of lesion mass and calcification,the reconstructed images suffer from blurring structures and increasing out-of-plane artifacts,which would directly influence the clinical diagnosis and treatment.Meanwhile,due to performance factors of the flat-panel detectors used in DBT imaging equipment,the images often appear dark current phenomenon,uneven distribution of light field and response nonuniformity of pixel,which directly affect the quality of projection data,resulting in various artifacts such as ring artifacts,strip artifacts,and so on.Therefore,this paper focus on the geometric characteristics of DBT imaging technology to study the strategy of optimizing the projection data,including simulated digital breast phantoms,corrected system matrix and study the influence of projection view distribution on DBT images quality reconstructed from limited-angle projection.Meanwhile,for the defect of the data acquisition process by the flat panel detectors,we research the impact of the projection data correction on the DBT reconstructed image quality.On the whole,the main works of this paper can be summarized as follows:1.A set of three-dimensional digital breast phantoms was simulated and the traditional iterative reconstruction system matrix was corrected based on the geometric characteristics of DBT imaging.We generate limited angular sinogram data of digital phantom by forward projection algorithm and then reconstructed DBT images by Simultaneous Algebraic Reconstruction Technique(SART).The difference between reconstructed tomographic image and the original phantom is used to verify the accuracy of correction operator.2.Based on the existing imaging geometry and image reconstruction methods of DBT,we research the effect of projection view distribution on DBT imaging quality,including different angular ranges with same projection numbers and angular increments,different numbers of projection data with same angular ranges and angular increments,nonuniform angular distribution with same angular ranges and projection numbers.At last,we use PWLS iterative reconstruction method based on Huber’s prior to reconstruct the images.3.We improve the correction methods of projection data during the DBT image acquisition and study the data correction process collected by flat panel detector.Referring to data correction methods of traditional detector and combining the specific data acquisition method of DBT imaging process,the dark field correction,light field and response nonuniformity correction are proceeded.Then,the DBT image reconstruction is performed by the PWLS iterative reconstruction algorithm based on the TGV regularization term,Finally,we summarize the contribution of projection data correction process to the DBT imaging.