Impact of instrumentation on DaTSCAN imaging: how feasible is the concept of cross-systems correction factor?

The aim of this investigation was to study the impact of applying different camera emission computed tomography (ECT) systems on semi-quantitative analysis of the novel dopamine transporter (DaT) radioligand [123I]-ioflupane (DaTSCAN) and whether applying a system-dependent correction factor is feasible to allow a migration of a ''normal'' reference data set between systems.

Two triple-head systems, PrismXP-3000 equipped with high resolution Fan-Beam (PRISM) and Irix (Philips Medical Systems, Europe) equipped with high resolution parallel whole collimators (IRIX), were compared in phantom and patient studies. Acquisition and quantification parameters were standardized. The anthropometric striatal calibration phantom (RSD Inc., CA, USA) at rising striatum: background ratios between 2-11 was studied. Fifty-one consecutive patients (25 females, age 64.5+/-11.9 years, and 26 males, age 60.6+/-12.7 years) were randomized to both cameras for acquisition time typically beginning 3.5 to 4.5 h after i.v. injection of 123I-Joflupane at 2.06+/-0.47 MBq/kg. Striatal uptake of the radioligand was categorized as normal or abnormal, and abnormal images were further subdivided into 3 severity levels. Striatum: cerebellum ratio (SCr) was determined by applying a fixed circular or rectangular striatal and fuzzy cerebellar regions. The effect of introducing scatter compensation (Sc) on quantification and on SCr was investigated.

The regression coefficient (Rc) of SCr computed for both systems was close to identity (IRIX=0.999x (PRISM)+0.48 and 1.01x(PRISM)+0.86 for right and left striatum). Rc of SCr values in the phantom studies was closer to identity when Sc was added (IRIX=0.982x [PRISM]+0.724 with Sc vs 1.22x[PRISM]-0.32 without Sc). SCr values were higher for IRIX by 5-10%. Including Sc increased the percent recovery and the linear dynamic range for both systems; however, fan beam recovery decreased at low SCr values. The delineation of the striatal boundaries improved after applying Sc. Identifying the cerebellar region was easier with PRISM owing to significantly higher count statistics. Both systems performed equally well with respect to image quality. Visual scoring by 2 observers correlated significantly on the k-test (K: 0.883, T: 10, P<0.0001).

A simple linear correction is applicable to ''normal'' reference data set to migrate from one system to another; however, data acquisition and quantification parameters need to be standardized. Higher recovery values are dependent on system resolution. The spatial distribution and image quality of DaTSCAN on different high-resolution systems applying standardized acquisition and reconstruction protocols is less operator dependent and does not affect visual rating of striatal DaT loss.