Dosimetric impacts of breathing motion-compensated reconstruction in radioembolization
Résumé
Aim: Selective Internal Radiation Treatment (SIRT) is a local liver cancer treatment planned using a predictive dosimetry from 99cmTc-MAA SPECT images which may be impacted by artifacts due to breathing motion. The aim of this work is to assess such impacts on real patient data by using a motion compensated reconstruction method. Materials and Methods: Twenty-nine patients were treated for hepatocellular carcinoma (10), cholangiocarcinoma (3), metastasis of colorectal cancer (4) and others (12) for a total of 31 SIRT treatments. All 99mTc SPECT images were reconstructed with two methods: conventional OSEM (3D) and motion compensated [1] (3Dcomp), both with attenuation, scatter and PSF corrections. 3Dcomp estimates breathing motion from listmode data and compensates each projection during reconstruction, leading to an almost motion free 3D image that can be created for any breathing phase. Predictive dosimetry was computed with Monte Carlo simulation, assuming that the biodistribution of 99mTc is the same as of 90Y. Seven volumes of interest (VOI) were delineated: liver, lungs, tumor(s), perfused liver (PL), hepatic reserve (HR), healthy perfused liver (HPL) and healthy liver (HL). Percentages of absolute dose difference (PaDD) were computed between 3D and 3Dcomp reconstructions, as well as the lung shunt fractions (LSF) and tumor-to-normal liver ratios (TN). The magnitudes of liver and tumor motion between the extreme respiratory phases were estimated by local rigid registration. The activity to be injected was assessed for each reconstruction using a mono-compartmental method. Results: PaDD between 3D and 3Dcomp reconstructions were 0.88±1.5% (max 7.5%) for liver, 23.1±29.8% (132.2%) for lungs, 6.7±9.5% for tumors, 0.87±1.4% (7.2%) for PL, 4.0±3.4% for HR, 1.3±1.5% (6.9%) for HPL and 1.7±1.9% (8.4%) for HL. PaDD were close to 0 for tumors which had a volume superior to 122 mL. For others, it depends on patients. Percentages of absolute difference of LSF and TN between reconstructions were 25.4±36.0% (range, 0.3-162.7%) and 3.7±5.5% (0.0-25.3%) respectively. The range of movement of the liver was 9.5±2.7mm (3.4-16.8mm) and that of the tumors was 10.9±2.4mm (5.6-17.1mm). PaDD in the planned activity to inject were 0.78±1.5% (max, 7.9%). Conclusion: Motion compensated reconstruction is feasible with real data and shows that this correction may be necessary for some patients only, with low-volume lesions or site-specific lesions, large motion amplitude. The criteria for its application have yet to be determined. References: [1] Robert et al. 2021. Data-driven motion compensated SPECT reconstruction for liver radioembolization. A Robert, S Rit, J Jomier, D Sarrut. Proceedings of the 16th International Meeting on Fully Three-Dimensional Image Reconstruction in Radiology and Nuclear Medicine 2021, 81-84, 2021