Optimal contrast as a function of noise for Butterworth filtering of 111In-pentetreotide SPECT when using model-based compensation

In ¹¹¹In-pentetreotide SPECT, it can be difficult to detect small tumors because of low spatial resolution and high noise levels. For an efficient cancer treatment, it is however important to detect the tumors at an early stage. The aim of this study was to perform optimization of Butterworth post-filtering of ¹¹¹In-pentetreotide SPECT, with regard to small tumor detection in the liver. All images were reconstructed with model-based OSEM reconstruction. Two collimators are evaluated, the extended low-energy general-purpose (ELEGP) and the medium-energy general-purpose (MEGP) collimator. The raw-data projection images are produced using Monte Carlo simulations of an anthropomorphic phantom with realistic ¹¹¹In-pentetreotide uptake, including seven spherical tumors in the liver. Evaluation was performed using tumor contrast as a function of background noise. According to the results, the detection of the smallest tumors is facilitated by not applying a filter. For somewhat larger tumors, a Butterworth filter of critical frequencies of 0.5-0.6 cm^-1 and an order of 10-12 is more appropriate. ELEGP proved to be better than MEGP for small tumor detection. The study is a continuation of a previous project where collimator choice, acquisition time, number of projection angles and OSEM settings were studied for the same geometry.