Planar and tomographic (SPECT) imaging of small volume targets using a Cross-Slit collimator

In vivo imaging techniques applied to small animals are important tools in basic research. Scintigraphic planar and tomographic images allow in vivo functional expression of the protocol's effect on the animal's organs to be assessed. In that way, long term dynamical studies can be developed on the same animal, which can be used as its own control, helping to reduce statistical uncertainties, to cut costs and to improve ethical aspects of the experiment. Considering the small size of those animals' organs, specially adapted instruments must be used in order to reach, simultaneously, good spatial resolution and sensitivity. Because of the intrinsic magnification factor, small-size pinhole-like collimators seem to be the obvious option to reach good spatial resolution. However, sensitivity is limited by the aperture's size. In this work, we propose the use of a combination of two 11-mm long, 0.8-mm width, mutually perpendicular slits as the collimator, in combination with appropriate restoration algorithms. By being narrow, information on high resolution is encoded on the recorded image, while, by being long, a large effective open area allows for good sensitivity. The slits are short enough such that individual open elements produce almost identical projections of the target. In a pre-processing stage, an implementation of the Maximum Likelihood Expectation Maximization algorithm was used to decode individual projections, producing high resolution planar images of the target. The same algorithm was used to combine those pre-processed individual projections, this time as if they would be obtained with a small size single pinhole collimator, obtaining a three-dimensional (3D) model of the target's radiotracer distribution. Monte Carlo simulations and physical phantoms were used to test the proposed methodology. Spatial resolution, equivalent to the one obtained with a 1.5-mm diameter pinhole collimator, was reached. Considering the collimator's open area, sensitivity should increase by a factor nearly 9, when compared with the same 1.5-mm diameter pinhole collimator. The experimental results are consistent with that theoretical value. Additionally, the pre-processing stage does not introduces a significant increase in the total processing time necessary for convergence to the reconstructed 3D object model. In this way, this kind of collimator appears as a good alternative to reduce the acquisition time, improving ethical aspects on animal management, as well as to allow the 3D target emission model to be obtained with no increased machine burden.