Quantitative accuracy of dopaminergic neurotransmission imaging with (123)I SPECT.

UNLABELLED 123I-Labeled radiotracers are suitable for in vivo imaging of the dopaminergic system by SPECT. However, precise measurement of striatal uptake is limited by scatter, attenuation, and the finite spatial resolution of the camera. We studied the quantitative accuracy that can be achieved with (123)I SPECT of the dopaminergic neurotransmission system. METHODS Using a Monte Carlo simulation and brain phantom experiments, we studied the biases in brain and striatal absolute uptake estimates and in binding potential (BP) values for different processing schemes with corrections for attenuation, scatter, and the partial-volume effect. RESULTS Without any correction, brain activity was underestimated by at least 65%, and absolute striatal activity measured in regions corresponding to the anatomic contours of the striata was underestimated by about 90%. With scatter and attenuation corrections only, estimated brain activity was accurate within 10%; however, striatal activity remained underestimated by about 50%, and BP values were underestimated by more than 50%. When combined with attenuation and scatter corrections, anatomically guided partial-volume effect correction (PVC) reduced the biases in striatal activity estimates and in BP values to about 10%. PVC reliability was affected by errors in registering SPECT with anatomic images, in segmenting anatomic images, and in estimating the spatial resolution. With registration errors of 1 voxel (2.1 x 2.1 x 3.6 mm(3)) in all directions and of 15 degrees around the axial direction, PVC still improved the accuracy of striatal activity and BP estimates compared with scatter and attenuation corrections alone, the errors being within 25%. A 50% overestimation of the striatal volume yielded an approximate 30% change in striatal activity estimates with respect to no overestimation but still provided striatal activity estimates that were more accurate than those obtained without PVC (average errors +/- 1 SD were -22.5% +/- 1.0% with PVC and -49.0% +/- 5.5% without PVC). A 2-mm error in the spatial resolution estimate changed the striatal activity and BP estimates by no more than 10%. CONCLUSION Accurate estimates of striatal uptake and BP in (123)I brain SPECT are feasible with PVC, even with small errors in registering SPECT with anatomic data or in segmenting the striata.