A fillable micro-hollow sphere lesion detection phantom using superposition.

ABSTRACT

The lesion detection performance of SPECT and PET scanners is most commonly evaluated with a phantom containing hollow spheres in a background chamber at a specified radionuclide contrast ratio. However, there are limitations associated with a miniature version of a hollow sphere phantom for small-animal SPECT and PET scanners. One issue is that the 'wall effect' associated with zero activity in the sphere wall and fill port causes significant errors for small diameter spheres. Another issue is that there are practical difficulties in fabricating and in filling very small spheres (<3 mm diameter). The need for lesion detection performance assessment of small-animal scanners has motivated our development of a micro-hollow sphere phantom that utilizes the principle of superposition. The phantom is fabricated by stereolithography and has interchangeable sectors containing hollow spheres with volumes ranging from 1 to 14 microL (diameters ranging from 1.25 to 3.0 mm). A simple 60 degrees internal rotation switches the positions of three such sectors with their corresponding background regions. Raw data from scans of each rotated configuration are combined and reconstructed to yield superposition images. Since the sphere counts and background counts are acquired separately, the wall effect is eliminated. The raw data are subsampled randomly prior to summation and reconstruction to specify the desired sphere-to-background contrast ratio of the superposition image. A set of images with multiple contrast ratios is generated for visual assessment of lesion detection thresholds. To demonstrate the utility of the phantom, data were acquired with a multi-pinhole SPECT/CT scanner. Micro-liter syringes were successful in filling the small hollow spheres, and the accuracy of the dispensed volume was validated through repeated filling and weighing of the spheres. The phantom's internal rotation and the data analysis process were successful in producing the expected superposition images. Visual inspection of the multi-contrast images provided simple determination of lesion detection thresholds for this scanner (41 ratio for 1.5 mm spheres and 31 ratio for 2.0 mm spheres) at a specified cumulated background concentration (30 kBq-min microL(-1)). In summary, the micro-hollow sphere phantom demonstrated its practical utility for lesion detection evaluation and is well suited for comparing the task-based performance of small-animal SPECT and PET scanners.