Digital breast tomosynthesis reconstruction using spatially weighted non-convex regularization

Regularization is an effective strategy for reducing noise in tomographic reconstruction. This paper proposes a spatially weighted non-convex (SWNC) regularization method for digital breast tomosynthesis (DBT) image reconstruction. With a non-convex cost function, this method can suppress noise without blurring microcalcifications (MC) and spiculations of masses. To minimize the non-convex cost function, we apply a majorize-minimize separable quadratic surrogate algorithm (MM-SQS) that is further accelerated by ordered subsets (OS). We applied the new method to a heterogeneous breast phantom and to human subject DBT data, and observed improved image quality in both situations. A quantitative study also showed that the SWNC method can significantly enhance the contrast-to-noise ratio of MCs. By properly selecting its parameters, the SWNC regularizer can preserve the appearance of the mass margins and breast parenchyma.

[1]  Mathews Jacob,et al.  Nonlocal Regularization of Inverse Problems: A Unified Variational Framework , 2013, IEEE Transactions on Image Processing.

[2]  Lubomir M. Hadjiiski,et al.  Digital breast tomosynthesis: computer-aided detection of clustered microcalcifications on planar projection images , 2014, Physics in medicine and biology.

[3]  A. Kak,et al.  Simultaneous Algebraic Reconstruction Technique (SART): A Superior Implementation of the Art Algorithm , 1984, Ultrasonic imaging.

[4]  Lubomir M. Hadjiiski,et al.  Multiscale bilateral filtering for improving image quality in digital breast tomosynthesis. , 2014, Medical physics.

[5]  Berkman Sahiner,et al.  Computer-aided detection of clustered microcalcifications in multiscale bilateral filtering regularized reconstructed digital breast tomosynthesis volume. , 2014, Medical physics.

[6]  Laurent D. Cohen,et al.  Non-local Regularization of Inverse Problems , 2008, ECCV.

[7]  Hakan Erdogan,et al.  Ordered subsets algorithms for transmission tomography. , 1999, Physics in medicine and biology.

[8]  Lubomir M. Hadjiiski,et al.  A comparative study of limited-angle cone-beam reconstruction methods for breast tomosynthesis. , 2006, Medical physics.

[9]  Jeffrey A. Fessler,et al.  Accelerating Ordered Subsets Image Reconstruction for X-ray CT Using Spatially Nonuniform Optimization Transfer , 2013, IEEE Transactions on Medical Imaging.

[10]  D. Hunter,et al.  Quantile Regression via an MM Algorithm , 2000 .

[11]  Heang-Ping Chan,et al.  Computer-aided detection system for clustered microcalcifications in digital breast tomosynthesis using joint information from volumetric and planar projection images , 2015, Physics in medicine and biology.

[12]  Kris Thielemans,et al.  Object dependency of resolution in reconstruction algorithms with interiteration filtering applied to PET data , 2004, IEEE Transactions on Medical Imaging.

[13]  Heang-Ping Chan,et al.  Digital breast tomosynthesis: observer performance of clustered microcalcification detection on breast phantom images acquired with an experimental system using variable scan angles, angular increments, and number of projection views. , 2014, Radiology.

[14]  Guobao Wang,et al.  Patch-based regularization for iterative PET image reconstruction , 2011, 2011 IEEE International Symposium on Biomedical Imaging: From Nano to Macro.