A Gelatin Liver Phantom of Suspended ⁹⁰Y Resin Microspheres to Simulate the Physiologic Microsphere Biodistribution of a Postradioembolization Liver

Phantom Construction

Institutional Review Board approval was not required for the conduct and publication of this phantom study, which did not involve any human subjects. The phantom design was inspired by a gel-based phantom used by Goedicke et al. to overcome the problem of microsphere sedimentation (7). For this phantom, the medium used to suspend the microspheres was gelatin. A 5%-by-weight solution of gelatin was prepared by slowly dissolving 75 g of gelatin into 1,500 mL of distilled sterile water (for injection/irrigation). The solution was kept at 45°C on a hot plate and magnetically stirred at 300 rpm to ensure that the gelatin would not set. The liver phantom was a 1,330-mL rectangular homeware container made of clear, rigid plastic. The 2 tumor inserts were round containers made of similar plastic material, measuring 27 and 58 mL by volume. A magnetic stir bar (Spinbar; Sigma-Aldrich) was placed into each of the 3 inserts, and then the inserts were filled to maximum capacity with gelatin. The inserts were kept on a hot plate with the stir bar at 300 rpm to ensure proper distribution of heat and to prevent setting.

⁹⁰Y resin microspheres from a full vial with a total activity of 3.144 GBq were carefully dispensed into each of the 3 inserts to achieve a realistic tumor–to–normal liver ratio of approximately 5 for hepatocellular carcinoma (10). The total vial activity was determined from calibration factors provided by the manufacturer, to an uncertainty of ±10%. For this vial, the derived total activity of 3.144 ± 0.31 GBq was in good agreement with the measured activity of 3.139 GBq by the dose calibrator (CRC-35; Capintec) using the ⁹⁰Y setting of 480 × 10. Individual activities were 281 ± 28 MBq in 27 mL (10.4 MBq/mL) and 573 ± 57 MBq in 58 mL (9.9 MBq/mL) for the 2 tumor inserts and 2,290 ± 229 MBq in 1,245 mL (1.8 MBq/mL) for the liver insert. All ⁹⁰Y activities were determined by volume from the original vial of well-suspended microspheres. All microspheres were completely dispensed into the phantom, with negligible residual vial activity.

After the microspheres were added, the hot plate was turned off and the gelatin was rapidly cooled in an ice-water bath while still being stirred. This stirring process resulted in a gelatin-based suspension of heterogeneously distributed microspheres throughout both tumor and nontumor liver compartments. The magnetic stir bars were intentionally left within the inserts to simulate the metallic coils used for prophylactic coil embolization. The plastic lids of all 3 inserts were secured with adhesive tape. The 2 tumor inserts were placed into the liver insert to complete the liver phantom. Therefore, the whole-liver total activity was 3.144 ± 0.31 GBq distributed throughout 1,330 mL of gelatin, with a mean radioconcentration of 2.36 MBq/mL. To simulate the attenuation within a patient, the liver phantom was fixed using adhesive strips near the bottom of a 5-liter plastic body phantom and further weighed down using saline bags placed on its lid. The body phantom was then completely filled with water and sealed with adhesive tape.

⁹⁰Y PET/CT Protocol

⁹⁰Y PET/CT scans were obtained on a Biograph mCT-S(64)4R (Siemens Medical Solutions), which uses lutetium oxyorthosilicate crystals and has time-of-flight capability. Because ⁹⁰Y was not available on this system as a radionuclide choice at the time of this study, all studies were acquired using ⁸⁶Y settings, with a half-life of 14.74 h and a branching ratio of 0.33. PET was acquired at 15 min per bed position and reconstructed using the vendor-supplied TrueX algorithm plus time-of-flight reconstruction (UltraHD-PET; Siemens) with 1 iteration, 21 subsets, and a filter of 8 mm, decay-corrected to the start of the scan. CT was performed at 120 kVp, 50 mAs (CARE Dose 4D; Siemens), a Z-coverage per rotation of 64 × 0.6 mm, 5-mm slices, and a pitch of 0.75. All phantom scans were obtained using a single bed position. All reconstructed ⁹⁰Y PET/CT images were qualitatively reviewed; ⁹⁰Y PET quantification of activity was not performed.

Phantom Physical Density

The main constituent of gelatin is collagen. Because our liver phantom was prepared in a nonsterile manner, environmental bacteria or fungi may have been inadvertently introduced into the phantom, with the potential to cause gelatin hydrolysis. Furthermore, the effect of ⁹⁰Y β radiation on the structural integrity of gelatin has not been well described. For our study, the physical density (g/cm³) of the liver phantom was empirically used as a surrogate measure of possible gelatin breakdown or leakage occurring during the study. We postulated that as gelatin degrades, its density may change due to the upward migration of air pockets trapped during preparation of the liver phantom.

To avoid radiation exposure to the operator, the physical density of gelatin was indirectly measured based on the CT component of the PET/CT scan. By assuming a linear relationship between the CT number (Hounsfield unit) and the physical density of a water-based object of interest (11), we calculated the physical density of the liver phantom as (mean CT number + 1,000)/1,000. CT analysis was performed using OsiriX software (version 5.6; Pixmeo). The mean CT number of the liver phantom was determined from a cylindric volume of interest with a 4-cm diameter and 4.5-cm length, placed within a region of gelatin free from large air pockets or tumor inserts. This volume of interest was repeated across all phantom scans to obtain its mean physical density and compared with an expected result of 1.05 g/cm³. As a control, the physical density of water within the body phantom was similarly measured and compared with an expected result of 1 g/cm³.