Loading...

Simulation and Acceleration of Image Reconstruction for a Typical PET Imaging System Using GPU

Sadat Shahabi, Mohsen | 2015

741 Viewed
  1. Type of Document: M.Sc. Thesis
  2. Language: Farsi
  3. Document No: 46880 (46)
  4. University: Sharif University of Technology
  5. Department: Energy Engineering
  6. Advisor(s): Vossoughi, Nasser; Vosughi Vahdat, Bijan
  7. Abstract:
  8. Positron Emmision Tomography is one of the most important methods in molecular imaging field. One of the challenges over PET imaging is the resolution of reconstructed images. To overcome this problem, one can use more exact image reconstruction methods called, iterative methods. It has been showed that Iterative methods like MLEM have advantages over Analytical ones. Main drawback of iterative methods is time consuming computation. Then you should reduce the image reconstruction’s time for clinical time constraints. Using Graphic Proccessing Unit for accelerating computational works, raised in recent years. In this work, we used GPU for accelerating pet image reconstruction and simulation using GATE software. For this purpose, we simulated GE Discovery STE system and used four different phantoms. We perfomed the image reconstruction of PET system by MLEM method. We used matlab and CUDA for implementing this reconstruction method. The GPU that we used was GTS450 from NVIDIA corporation. First of all, We evaluated simulation results by comparing them with the results of reffrence papers. Then we implemented GATE simulations on GPU and gained speed up factor about 20. In the image reconstruction step, we used ASCII output of GATE and reconstructed it by MATLAB. Then evaluated image reconstruction results by STIRv3.0. Finaly we used CUDA to implement image reconstruction on GPU. We gaind speed up factor about 5.5. Then using GPU for accelerating image reconstruction could keep the more exact image reconstructions, still reliable for clinical times
  9. Keywords:
  10. Image Reconstruction ; Positron Emission Tomography (PET) ; Graphic Processing ; Computed Unified Device Architecture (CUDA)Platform ; Geant4 Application for Tomographic (GATE)Emission ; Maximum Likelihood-Expectation Maximization (MLEM)

 Digital Object List

 Bookmark

No TOC