Development of a scatter correction technique for planar 99mTc-MAA imaging to improve accuracy in lung shunt fraction estimation

Purpose: Prior to Y-90 selective internal radiation therapy (SIRT) treatment, Tc-99m-MAA scintigraphy imaging is used in the estimation of the lung shunt fraction (LSF). Planar imaging is recommended for determining a LSF ratio. However, the estimate may be affected by scatter contributions, attenuation and respiratory motion. The objective of this study was to correct for the effects of scatter in the LSF, towards the determination of a more accurate estimation method of LSF derived from planar scintigraphy imaging, which is recommended by international guidelines. Methods: The open access SIMIND Monte Carlo modelling software was used to estimate an optimum scatter window (SW) for scatter correction. The uncertainties associated with scatter and scatter contributions from the liver on the LSF were evaluated using an anthropomorphic thorax phantom and a virtual Vox-Man phantom. A brief retrospective examination of patient scans and tumour location investigated the impact that the inclusion of the simulated scatter corrections had on the LSF estimation. Results: The percentage overestimation of the manufacturer recommended method of LSF estimation was 192%. SW corrections improved the uncertainty to within 19% for the range of known LSFs. Similar findings were observed for our patient and tumour location studies. Conclusion: The incorporated scatter corrections can significantly improve the accuracy of the LSF estimation, thereby providing a robust gamma camera, patient and tumour depth specific correction which is easily implementable. This is supported by Monte Carlo, phantom and preliminary patient studies.

Automated summary

The study corrected for scatter effects in LSF estimation using Monte Carlo simulation software, resulting in improved accuracy and reduced overestimation. Scatter window corrections were effective in reducing uncertainty and improving LSF estimation within a range of known values.