Signal to Noise Ratio optimization for extended sources with a new kind of MURA masks

Abstract

Using coded aperture, for localization of radioactive hot-spots, results in enhanced efficiency and under certain configurations wide Field of View (FOV). We present a coded aperture assembly technique which can be restructured easily, as well as the reduction of the intrinsic noise of coded apertures constructed with this technique, when they localize spatially extended γ-emitters. Specifically, Modified-Uniformly-Redundant-Array (MURA) coded apertures are structured by embedding lead spheres in a matrix of positions machined inside a transparent medium such as acrylic glass, resulting in an advantageous transparent to opaque area ratio and thus an improved detection efficiency. This configuration also induces a systematic, element-wise, noise on the Point-Spread-Function (PSF) of the correlation matrix. When imaging with these apertures extended hot-spots, a penumbra phenomenon occurrs and reduces this intrinsic noise, in the way a kernel filter does. Fast-Fourier-Transform (FFT) is used to analyze the effect of this phenomenon on the correlation matrix and to explain the maximization of its Signal-to-Noise Ratio (SNR) for certain extent of the hot-spots. Simulations have been used for the detailed study of the SNR dependence on the dimensions of the hot-spot, while experiments with two 99mTc cylindrical sources with 11mm and 24mm diameter, respectively and 1.5 MBq activity each, confirm the reduction of the intrinsic noise. The results define the way of optimization of the imaging setup for the detection of extended hot-spots. Such an optimization could be useful for example in the case of lymph nodes or thyroid remnant imaging in nuclear medicine. Finally, we propose a kernel filter, derived by the Auto-Correlation-Function (ACF), to be applied on PSFs with high intrinsic noise, in order to eliminate it.