Developing a 4-D Dynamic Model for GammaTile Dose Distribution

Abstract

GammaTile consists of Cesium-131 seeds imbedded in square 2cm tiles for intracranial resectable neoplasms. Currently, intra- and inter-tile source motion as the resection cavity evolves is not well characterized. We hypothesize that source motion is influenced by the overall tile motion which can be elucidated by 4D models of seed motion. Our patient underwent 23 fractions of EBRT after GammaTile for glioblastoma as a combined modality approach. Individual sources were identified and assigned to tiles on the post-operative, daily cone beam, and post-radiotherapy CT scans. All imaging was registered to the day 1 post-op/implant CT to utilize the same 3D space for tracking source movement. A programming environment was used to model source distances and generate dose clouds based on the seed position to create a 4D isodose volume. These dose clouds were compared to those calculated from the day 1 post-implant static isodose volume. The statistical analysis was performed using JMP®. We compared the 4D vs. static isodose volumes. The DICE score for each isodose volume was evaluated (for 30, 60, and 90 Gy the DICE scores were 0.95, 0.94, and 0.92 respectively). We found a steady decline in DICE score as the isodose line increased: DICE = -0.00065*Dose (Gy) + 0.976. Looking at the distance moved, tiles moved a mean of 5.8 mm (1.3-9.6) and appeared to remain adherent to the resection cavity wall in the first 81 days. Looking at the relative motion of the seeds within a tile, we see the tile retains a planar shape during the first 81 days and seeds move a mean of 2.2 mm (1.8-2.5) towards the center of the tile, which was a 50% contraction of the tile area. Table 1 details changes in the mean distance of source from tile center and volume occupied by all sources over time. This modeling of the tile movement/contraction is the first reported proof of concept in an approach relating tile motion with respect to sources and cavity over time. Tiles moved a mean of 5.8 mm and shrunk in the first 81 days. Discerning both how surgical cavity healing impacts seed/tile migration and how deformation impacts the dose distribution of the radiation delivered may provide a more accurate estimate of dose received by normal structures and at-risk tumor bed.