Performance Evaluation of FBP Reconstruction in SPECT Imaging

Document Type: Original Paper

Authors

1 Ph.D. Student in Medical Physics, Tarbiat Modarres University, Tehran, Iran.

2 Assistant Professor, Medical Physics Dept., Tarbiat Modarres University, Tehran, Iran.

3 M.D., Dept. of Nuclear Medicine, Rajaei Heart Centere, Tehran, Iran.

4 Associate Professor, Dept. of Biostatistics, Tarbiat Modarres University, Tehran, Iran.

Abstract

Introduction:  The  purpose  of  this  study  is  to  define  the  optimal  parameters  for  the  tomographic 
reconstruction procedure in a routine single photon emission tomography. The Hoffman brain phantom 
is modified to evaluate the reconstruction method. The phantom was imaged in a 3 and 2-dimensional 
conformation and the results were compared.  
Materials  and  Methods:  The  2D  phantom  (slices)  was  imaged  independently.  The  tomographic 
image was performed using the phantom in 3D conformation. Planar image was used as the reference 
image for each corresponding reconstructed slice image. A set of projection data with different count 
densities were obtained from 99mTc with the phantom. The data were transformed into interfile format 
and transferred to personal computer. A software was designed in Matlab (Version 7) to support the 
phantom  and  the  automatic  calculation.  All  the  data  were  reconstructed  using  back-projection 
technique with the ramp, Shepp-Logan, Cosine, Hamming, Hanning (each filter having so different 
parameters), Butterworth, Metz and Wiener (each filter having 2000 different parameters). The quality 
of the images was estimated by comparing the reconstructed images with the reference image by the 
use of universal image quality index (UIQI). 
Results: All the filters tested score substantially better than the control (P<0.05). The data from this 
study suggests that the Hanning and Hamming are the weakest of the seven filter types tested (mean of 
UIQI=  0.2218,  0.2219).  This  is  due  to  response  of  Hanning  and  Hamming  filters  at  the  low 
frequencies.  The  quantitative  results  of  this  study  indicate  that  Wiener  filter  provided  best  image 
quality of the phantom than the other filters regardless of which collimator is used. The mean of UIQI 
was the most for wiener filter in comparison with the other filters (P<0.05).  
Discussion and Conclusion: These results clearly show that Wiener filter improves quality image and 
the extent of the improvement has been quantified. The optimum parameters obtained for the filters 
show direct relation with total counts in the study. 

Main Subjects