Preliminary Studies on the Feasibility of Addition of Vertex View to Conventional Brain SPECT

Abstract

We have investigated the improvement in resolution and sensitivity for brain imaging which would result by the addition of a single stationary vertex view to the tomographic data. This method has the practical advantage of being relatively inexpensive and easy to implement. The uniform Cramer Rao bound is a plot of the minimum achievable standard deviation for estimating the pixel intensity as a function of the bias gradient length. Uniform CR bound analysis indicated an improvement in performance when the vertex detector is added, especially for centrally located pixels for which improvement is seen over the useful depth for brain imaging. Simulation experiments were done with a simple six slice phantom and with the Hoffman brain phantom. Visual inspection of the reconstructed images showed improved resolution and noise characteristics over reconstructed images without the vertex data. Quantitatively, substantial reduction in mean square error was observed for a plane close to the vertex detector. Improvement reduced as distance from the vertex detector is increased. Background activities inside the field of view of the vertex detector but not the tomograph were represented by several blobs of activity on a plane lying outside the reconstruction volume. This activity was estimated by 3D spline fitting jointly with the image reconstruction process. Adding the vertex view to conventional brain SPECT should lead to improved cortical imaging, and to moderate improvement for deep structures.