Title   name
2007년 KSIAM 봄 학술대회
  Speaker   Lee, Nam-Yong  
  Date 2007-05-25
  Place KAIST
  File  의 1 번째 Real Media 동영상입니다. 의 1 번째 강연자료입니다.
Abstract : The projection data of the emission activity of interests is blurred by the point response functionof the collimator that is used to define the range of directions where photons can be detected. Thepoint response function of the collimator is not spatially invariant, but it depends on the distancebetween the point and the collimator. Conventional Fourier correction methods limit deblurringability due to the simplification of the Fourier description, and are weak against to noise, sincethe deblurring process follows an inverse filtering. In this work we use ML-EM correctionmethod to reconstruct images with efficient removal of the distance dependent blurring. Wecompute the projector and the backprojector as a sum of convolutions with distance-dependentpoint response functions instead of matrix form. We conducted several simulation studies tocompare the performance of the proposed method with that of Fourier correction methods. Theresult shows that the proposed method outperforms Fourier correction methods objectively andsubjectively.