Related Experiment Video
Updated: Aug 5, 2026

Measuring 3D In-vivo Shoulder Kinematics using Biplanar Videoradiography
Published on: March 12, 2021
DBT reconstruction based on 3D coordinate transformation of multi-angle projections
1Nanjing University of Posts and Telecommunications, College of Electronic and Optical Engineering & College of Flexible Electronics (Future Technology), Nanjing 210023, China.
Abstract:
PurposeDigital Breast Tomosynthesis (DBT) is an emerging imaging technique for diagnosing breast disease. It reconstructs three-dimensional (3D) tomographic images from a limited number of projections within a restricted angular range. This technique effectively reduces tissue overlap and improves lesion detection rates. However, conventional reconstruction relies on a paraxial approximation, neglecting the increasing vertical geometric distortion at the detector periphery. This, distortion, blurring the microcalcifications and subtle lesions, is to be decreased by a proposed reconstruction.ApproachThe proposed reconstruction uses 3D coordinate transformation to overcome the geometric distortion and so it follows a sequential five-step pipeline: preprocessing, coordinate transformation, interpolation, weighting, and reconstructing. First, multi-angle projections undergo the preprocessing of background segmentation and filtering. Second, a 3D coordinate transformation is applied to each projection based on its acquisition angle and depth. Then, linear interpolation is used to fill data gaps caused by the spatial shift. Next, an angular weighting scheme balances the sampling density from different projections. Finally, tomographic images are reconstructed from these processed projections.ResultsExperimental results show that the 3D coordinate transformation improves geometric accuracy and reduces distortion. Specifically, for FBP reconstruction, the proposed method increased Peak Signal-to-Noise Ratio (PSNR) by 1.19 dB and reduced Standard Deviation (SD) by 403.53. For MLEM reconstruction, the Contrast-to-Noise Ratio (CNR) is improved by approximately 15.4%. Clinical data further confirm that these quantitative improvements help in differentiating lesions from the surrounding background.ConclusionsThe proposed DBT reconstruction based on 3D coordinate transformation improves image quality, showing potential for broader clinical applications.

