Lesion detection is a common task in PET and SPECT cancer imaging studies. To assess scanner performance for lesion detection, phantoms with fillable hollow spheres are typically used, both for research studies and in industry standards (such as NEMA NU-2). One of the major and well-known limitations of hollow spheres phantoms is the """"""""dead volume"""""""" associated with non-zero activity surrounding the active volume of the sphere, namely due to the sphere wall, mounting stem, and air bubbles. Because of the dead volume, a fillable sphere does not accurately represent the true clinical situation of a cancerous lesion in a background region of lesser activity concentration. For example, a hollow sphere with 10 mm internal diameter and 1.0 mm wall thickness has a wall volume that is 73% of the internal sphere volume. Our preliminary data have shown that the dead volume from a sphere of this size causes the lesion-to-background ratio to be underestimated significantly on a clinical PET scanner. Now that ultra-high resolution micro-PET and micro-SPECT scanners are commercially available, there is a need for accurate nuclear imaging phantoms with very small spheres. However, conventional phantoms are not feasible since the dead volume becomes dominant for smaller spheres, due to the cubic relationship of the wall thickness and volume. We therefore propose to develop a new variety of lesion detection phantoms that eliminate the errors from surrounding regions of non-zero activity while also being practical to use. This new design is based on the principle of superposition, where a pair of separate acquisitions is performed and the sinogram data are summed. In order for this approach to be valid, the physics underlying the two scans must be constant. Our proposed phantoms are designed to fulfill this requirement and will yield accurate results.
Our aims are to develop and test a prototype full-size phantom for clinical scanners (PET and SPECT) and also a miniature version for small animal scanners (micro-PET and micro-SPECT) having spheres as small as 1.5 mm inside diameter. We will perform quantitative imaging tests to evaluate the improved accuracy in comparison to conventional phantoms. The goal of this project is to develop and test novel phantoms for performance evaluation of lesion detection in PET and SPECT oncology imaging, for both clinical scanners and pre-clinical small animal scanners. Conventional fillable spheres phantoms have a well-known limitation in that the sphere wall creates a region of zero activity (""""""""dead volume"""""""") that significantly affects assessment of lesion detection. The proposed new design eliminates the dead volume problem and will improve clinicians' and researchers' understanding of the lesion detection capabilities of PET and SPECT scanners. ? ? ?
| Difilippo, Frank P; Gallo, Sven L; Klatte, Ryan S et al. (2010) A fillable micro-hollow sphere lesion detection phantom using superposition. Phys Med Biol 55:5363-81 |
