Optimal collimator design using Monte Carlo Simulation and RSM for breast scintigraphy

Abstract

Scintimammography can be a useful adjunct to physical examination andmammography for the detection and characterization of breast tumors, especially forpatients with dense breast tissue, architectural distortion of the breast, breast implant or with equivocal mammography. However, one major limitation of Scintimammography isits poor sensitivity and image quality for small lesions (<1.5cm). The aim of this work isto optimize the collimator parameters of a Scintimammography system to achieve betterimage quality by using Monte Carlo Simulation and the Response Surface Method. Two software packages have been used for this purpose: SIMIND is a Monte CarloSimulation program developed by Dr. Michael Ljungberg. NCSS is a statistical analysissoftware package used for the Response Surface Method (RSM). Monte Carlo has awide usage in nuclear medicine imaging, however RSM has not been used much in this area. RSM stems from science disciplines in which experiments are performed to studythe unknown relation between a set of variables and the system output, or response. Weran SIMIND to simulate a planar gamma camera system and carry out the experiments.NCSS was used for the optimization process. The SNR (signal-to-noise ratio) is selectedas the detectability index. A MATLAB program was written to compute SNR valuesfrom the outputs of SIMIND. The breast was modeled as a cylinder full of water and aspherical lesion with a diameter of 0.3 cm. Our results show that for a constant septa sizeof 0.02 cm, we can obtain optimum detectability when the diameter is 0.132 cm and collimator length is 1.449.|Keywords: Scintimammography, Monte Carlo simulations, Response Surface Methods, Collimator Optimization.