Evaluation of the maximum a posteriori (MAP) reconstruction on a microPET Focus220 scanner

Abstract

Spatial resolution and noise are strongly related properties and therefore meaningful evaluations of imaging systems and reconstruction algorithms should always include estimates of both. We have evaluated the performance of a statistical maximum a posteriori probability (MAP) reconstruction algorithm in terms of contrast recovery and signal to noise ratio (SNR) in comparison to filtered-backprojection (FBP) on a microPET Focus220 scanner. Printed <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">18</sup> F point sources sitting on uniform backgrounds were used as contrast recovery phantoms to evaluate the MAP and FBP performance. With this inkjet printed source method, the contrast recovery phantom was created with well controlled contrast levels and no finite boundaries were introduced between signal and background. The activities and contrast levels we used in our experiment were close to tumor uptakes and tumor to background contrasts in realistic mouse scans. Contrast evaluation demonstrated that at any given noise level, MAP reconstruction performance was better than FBP. To evaluate the effects of scatter in the two image reconstruction algorithms, analytical scatter correction was applied on a small animal mouse phantom containing water and air filled cavities and then reconstructed with both FBP and MAP algorithm. Residual activity in non-radioactive regions after scatter correction was proved to be similar for MAP and FBP reconstructions.