Optimisation of reconstruction, volumetry and dosimetry for 99mTc-SPECT and 90Y-PET images: Towards reliable dose-volume histograms for selective internal radiation therapy with 90Y-microspheres

Abstract

Purpose: In Selective Internal Radiation Therapy (SIRT), ^99mTc-MAA SPECT images are commonly used to predict microspheres distribution but recent works used ⁹⁰Y-microspheres PET images. Nevertheless, evaluation of the predictive power of ^99mTc-MAA has been hampered by the lack of reliable comparisons between ^99mTc-SPECT and ⁹⁰Y-PET images. Our aim was to determine the "in situ" optimisation procedure in order to reliably compare ^99mTc-SPECT and ⁹⁰Y-PET images and achieve optimal personal dosimetry.

Methods: We acquired ^99mTc-SPECT/CT and ⁹⁰Y-PET/CT images of NEMA and Jaszczak phantoms. We found the best reconstruction parameters for quantification and for volume estimations. We determined adaptive threshold curves on the volumetric reconstruction. We copied the optimised volumes on the quantitative reconstruction, named here the "cross volumes" technique. Finally, we compared ^99mTc-SPECT and ⁹⁰Y-PET Dose Volume Histograms.

Results: Our "in situ" optimisation procedure decreased errors on volumes and quantification (from -44.2% and -15.8% to -3.4% and -3.28%, respectively, for the 26.5mL PET phantom sphere). Moreover, ^99mTc-SPECT and ⁹⁰Y-PET DVHs were equivalent only after the optimisation procedure (difference in mean dose <5% for the three biggest spheres).

Conclusions: This work showed that a preliminary "in situ" phantom study was necessary to optimise volumes and quantification of ^99mTc-SPECT and ⁹⁰Y-PET images and allowed to achieve a reliable comparison between patient treatment planning and post implant dosimetry, notably by the use of the "cross volumes" technique. Methodology developed in this work will enable robust evaluations of the predictive power of ^99mTc-SPECT, as well as dose-response relationship and side effects in SIRT treatments.

Keywords: (90)Y-PET; SIRT DVHs; SIRT dosimetry; SIRT image optimisation.