Towards clinically useful coded apertures for planar nuclear medicine imaging.

Abstract

Coded apertures have the potential to increase imaging e ciency in nuclear medicine without degrading resolution, but near- eld artifacts are present even under idealised aperture and imaging conditions. The purpose of this work is to reduce artifacts prior to reconstruction, and to work towards coded apertures that are clinically useful. A ray-tracing simulator was developed. Far- eld conditions produce nearperfect images, but the simulation of distributed sources under idealised near- eld conditions results in the presence of artifacts. Three core concepts are introduced in this thesis: a novel rotatable array of identical limited- eld-of-view coded apertures; the use of high-resolution aperture projections; and the deliberate and counter-intuitive use of thin, highly transparent aperture material. An array of identical limited- eld-of-view coded apertures, which can be rotated so as to implement an existing artifact-reduction technique, was simulated. The artifacts that exist for a single coded aperture under idealised conditions are removed. This novel technique remains e ective when realistic near- eld conditions are introduced into the simulation. However, realistic apertures increase artifacts due to nite pinhole widths and nite thickness of the aperture material. To address the pinhole width problem, high-resolution patterns, in which the smallest hole corresponds to a projection area of 1 1 detector pixels, o er the best trade-o between e ciency and resolution despite the partial volume e ect. The nite aperture thickness problem is addressed by another novel concept; viz. the deliberate reduction in material thickness, which results in a highly transparent coded aperture. Simulation shows that this counter-intuitive approach diminishes collimation e ects. The implementation of any or all of these three core concepts, however, reduces count statistics. An ultra-near- eld geometry, which would ordinarily result in severe artifacts, can theoretically be used to maintain count statistics, without altering either patient dose or image acquisition time. This was veri ed by simulation. ii ABSTRACT A prior-state-of-the-art 1 mm thick tungsten coded aperture, and a deliberately highly transparent 100 m tungsten foil coded aperture, were constructed for use with a dual head gamma camera. Phantom studies of Technetium-99m point, line, syringe and printed distributed sources were performed. The experimental acquisitions veri ed the simulation results, for both the prior-state-of-the-art coded aperture and the novel high-transparency coded aperture. The results and arguments presented in this thesis point to the potential for these three core developments to produce high-quality coded aperture images in a fraction of the time that is taken for a collimator acquisition. The limiting factor appears to be the poor count statistics that result from the low sensitivity of current gamma cameras; a situation which looks set to change given current research trends.