CNR calculation methods.

Purpose

Prior to ⁹⁰Y radioembolization procedure, a pretherapy simulation using ^99mTc-MAA is performed. Alternatively, a small dosage of ⁹⁰Y microspheres could be used. We aimed to assess the accuracy of lung shunt fraction (LSF) estimation in both high activity ⁹⁰Y posttreatment and pretreatment scans with isotope activity of ~100 MBq, using different imaging techniques. Additionally, we assessed the feasibility of visualising hot and cold hepatic tumours in PET/CT and Bremsstrahlung SPECT/CT images.

Materials and methods

Anthropomorphic phantom including liver (with two spherical tumours) and lung inserts was filled with ⁹⁰Y chloride to simulate an LSF of 9.8%. The total initial activity in the liver was 1451 MBq, including 19.4 MBq in the hot sphere. Nine measurement sessions including PET/CT, SPECT/CT, and planar images were acquired at activities in the whole phantom ranging from 1618 MBq down to 43 MBq. The visibility of the tumours was appraised based on independent observers’ scores. Quantitatively, contrast-to-noise ratio (CNR) was calculated for both spheres in all images.

Results

LSF estimation. For high activity in the phantom, PET reconstructions slightly underestimated the LSF; absolute difference was <1.5pp (percent point). For activity <100 MBq, the LSF was overestimated. Both SPECT and planar scintigraphy overestimated the LSF for all activities.

Lesion visibility. For SPECT/CT, the cold tumour proved too small to be discernible (CNR <0.5) regardless of the ⁹⁰Y activity in the liver, while hot sphere was visible for activity >200 MBq (CNR>4). For PET/CT, the cold tumour was only visible with the highest ⁹⁰Y activity (CNR>4), whereas the hot one was seen for activity >100 MBq (CNR>5).

Conclusions

PET/CT may accurately estimate the LSF in a ⁹⁰Y posttreatment procedure. However, at low activities of about 100 MBq it seems to provide unreliable estimations. PET imaging provided better visualisation of both hot and cold tumours.