Estimation of lung shunt fraction from simultaneous fluoroscopic and nuclear images.

Objectives: Radioembolisation with yttrium-90 (⁹⁰Y) is increasingly used as a treatment of unresectable liver malignancies. For safety, scintigraphy of a scout dose of technetium-99m macroaggregated albumin (^99mTc-MAA) is performed prior to the delivery of the therapeutic dose to mimic the deposition of ⁹⁰Y. Procedures are currently limited by the necessary transfer of patients between the intervention room and the nuclear medicine department for scintigraphy. To cope with this limitation, a real time, interventional simultaneous fluoroscopic and nuclear imaging device, consisting of an X-ray c-arm combined with four gamma cameras, is currently being developed (1,2). This set-up enables simultaneous acquisition of fluoroscopic and nuclear images of the same field-of-view (FOV), thereby creating overlapping hybrid images in both the spatial and the temporal domain. The hybrid images provide a quick insight into the radionuclide distribution during the procedure. In addition, the accuracy of estimating lung shunt fraction (LSF) may be improved by handling breathing, attenuation and organ overlap effects. The purpose of the present simulation study was to evaluate the feasibility of estimating the LSF of the scout dose in the intervention room with this device.

Methods: Two XCAT phantoms, one male and one female, were used to simulate various LSFs (0, 1, 2, 5, 10, 15, 20, 25, 30 and 35%) resulting from a scout dose of 150 MBq ^99mTc-MAA. Hybrid images were Monte Carlo simulated for static breath hold (5 s) and dynamic breathing (10 frames of 0.5 s) acquisitions including all relevant effects. With the fluoroscopic X-ray image, nuclear images could be corrected for attenuation. In this, mean attenuation in the liver and lung regions of interest (ROIs) are estimated assuming a mono-energetic X-ray beam and homogeneous tissue composition in the lung and liver ROIs. Furthermore, LSF estimates were corrected for organ overlap effects by using a pre-treatment CT scan registered to the X-ray image. For comparison purposes, planar scintigraphy images (300 s acquisition time) - representing the current clinical standard - were also simulated. Estimated LSFs from the simulated projections were evaluated for all methods and compared to the phantom ground truth.

Results: Figure 1 shows static hybrid images of the male phantom with 20% LSF. In the clinically relevant range of 10-20% LSF, hybrid imaging, as well as planar scintigraphy, overestimated LSF with approximately 5 percentage points, with larger errors in the male phantom than in the female phantom. This may wrongfully deny treatment to eligible patients. Dynamic hybrid imaging consistently estimated LSF more accurate. The proposed attenuation correction algorithm improved estimated LSF to approximately 2 percentage points in the clinical relevant range. This error was further reduced to approximately 1 percentage point when LSF was also corrected for organ overlap.

Conclusion: Lung shunt fraction can be estimated more accurately with a simultaneous fluoroscopic and nuclear imaging device than with planar scintigraphy. The hybrid imaging device is capable of accurately estimating LSF within a few seconds in an interventional setting. Research Support: 1. Beijst C, Elschot M, Viergever MA, Jong HW de. Toward simultaneous real-time fluoroscopic and nuclear imaging in the intervention room. Radiology. 2015;278:232-238. 2. Velden S van der, Beijst C, Viergever MA, Jong HWA. de. Simultaneous fluoroscopic and nuclear imaging: Impact of collimator choice on nuclear image quality. Med Phys. 2016:in press. Figure 1: Simulated hybrid images of the male phantom with 20% LSF. (A) Static hybrid image of liver FOV, (B) static hybrid image of lung FOV. Liver delineations are indicated in red, lung delineations are indicated in green. Images are linearly window-levelled between 0 and 1.5 times the mean number of counts in the liver.

• Download figure
• Open in new tab
• Download powerpoint