Perelman School of Medicine at the University of Pennsylvania

Physics and Instrumentation Group

banner image

Image Reconstruction

(PI: Samuel Matej)

The main focus of our reconstruction work has been on research, development, and investigation of PET reconstruction methods and tools supporting qualitative and quantitative imaging. These approaches include tools for static imaging on novel whole-body scanners, as well as for static and dynamic reconstructions for scanners with specialized geometries developed in our group. Our works have been funded mostly by NIH research grants, but we also have a strong track-record of industry collaborations. Several reconstruction approaches developed within the group have been applied and successfully tested in clinical settings and are in routine use on commercial PET scanners (by Philips Medical Systems). Examples of such tools and approaches implemented on commercial scanners are: Fourier rebinning tools, iterative reconstruction approaches utilizing smooth spherically-symmetric basis functions (blobs) on efficient spatial grids, Row Action Maximum-Likelihood Algorithm (RAMLA), list-mode reconstruction tools (IRX), and Fourier-based analytic (fully-3D) reconstruction approach (3D-FRP).

Our recent works include development, implementation, and investigation of novel efficient Fourier-based TOF reconstruction approaches (both iterative and analytic) within the DIRECT framework (Direct Image Reconstruction for TOF data). We also have been expanding DIRECT reconstruction tools to accommodate new dynamic applications and novel PET instrumentation developed within the group, including approaches dealing with large axial gaps in the prototype PennPET Explorer scanner [link], resolution effects of the long-axial FOV scanners, spatial and temporal regularization approaches, and spatially variant resolution modeling reconstruction for the B-PET scanner [link] with limited angle data and strong angle-dependent and position-dependent depth of interaction and PSF deterioration effects. We also have been developing and investigating TOF-based approaches for CT-less simultaneous emission and attenuation reconstruction approaches (in conjunction with scatter estimation) for long-axial FOV scanner data.


Selected Publications

  1. Li Y, Defrise M, Matej S, Metzler SD. Fourier rebinning and consistency equations for time-of-flight PET planograms. Inverse Problems, 32, pp. 1-33, 2016. [Link]
  2. Li Y, Matej S, Metzler SD. A unified Fourier theory for time-of-flight PET data. Phys Med Biol, 61, pp. 601-624, 2016. [Link]
  3. Matej S, Li Y, Panetta J, Karp JS, Surti S. Image-based modeling of PSF deformation with application to limited angle PET data. IEEE Trans Nucl Sci, vol. 63, pp. 2599-2606, 2016. [Link]
  4. Matej S, Daube-Witherspoon ME, Karp JS. Analytic TOF PET reconstruction algorithm within DIRECT data partitioning framework. Phys Med Biol, vol. 61, pp. 3365-3386, 2016. Paper selected by PMB for their Highlights of 2016. [Link]