Analysis of spect/ct quantitation in soft tissue and lung density phantoms

We examine sources of uncertainty affecting quantitative performance of SPECT/CT imagery in objects with heterogeneous activity and attenuation (i.e. soft tissue and lung). Understanding the cause and amount of uncertainty in SPECT quantitation is a necessary step towards multicenter SPECT studies and for improving the consistency and standards of care. While quantitation has been studied using water filled phantoms, we seek to extend this effort to lung phantoms of varying densities spanning norm lung tissue (i.e., −750 HU) to diseased lung tissue (i.e., −950 HU). A series of images of a prototype lung phantom were acquired with a GE Discovery NM/CT 670. This phantom consists of a repeatable structure of two or three water-filled blocks and two expandable polyurethane foam (UEF) cylinders with activity mixed in the foam and one of the water blocks. Contributions of uncertainty due to VOI selection, intrinsic scanner repeatability, setup repeatability, phantom density, and scatter profile were examined. We found the effects of VOI selection, resolution spill-out, and density-dependent scatter-correction errors all have significant impact (> 5%) on quantitative bias. Scan-to-scan repeatability, even with small (∼ 1 cm) setup variation were typically small (< 2%).