Iterative reconstruction of fluorine-18 SPECT using geometric point response correction.

UNLABELLED This article demonstrates resolution recovery in 18F SPECT image reconstruction by using an iterative algorithm that corrects for the system geometric response. METHODS Patient and phantom studies were performed using a Picker PRISM 3000 three-detector SPECT system (Picker International, Inc., Cleveland, OH) to image 18F with 511 keV collimators. A measured point response function of the imaging system was used in an iterative reconstruction algorithm in which the projector and backprojector modeled the system point response function by using an efficient layer-by-layer blurring technique. The blurring function was a five-element kernel in the shape of a cross. The iterative reconstruction algorithm was an ordered-subset maximum-likelihood expectation maximization algorithm. RESULTS The iterative reconstruction algorithm with geometric response correction showed an improvement in resolution over the filtered backprojection reconstruction and the iterative reconstruction without correction. CONCLUSION The proposed iterative reconstruction algorithm with geometric response correction is efficient and effective with significant resolution recovery.