Room: AAPM ePoster Library
Intra-fractional tumour motion induced by respiration is a major source of uncertainty in radiotherapy. Using MLC-tracking can minimize this uncertainty by continuously reshaping the treatment beam according to the observed target motion. For large systematic shifts, however, this may lead to an under- or over-dosage due to the flattening-filter-free (FFF) beam profile. In this study, we demonstrate how real-time adaptive dose scaling during MLC-tracking can mitigate this effect on the Unity MR-linac (Elekta AB, Stockholm, Sweden).
MLC-tracking for IMRT was implemented on the Unity MR-linac. The Quasar MRI4D phantom (Modus Medical Devices, London ON) was used to generate Lujan motion (Cos4, T=5s, A=15mm). Additionally, the table was shifted by 2.5cm and 5cm to simulate large target shifts. The motion was estimated from 2D cine-MR (4Hz) using template matching. System latency (240-490ms) was compensated using a ridge regression prediction filter. The resulting position vector was used to shift the MLC aperture. The time-averaged target motion was extracted from the last two respiratory cycles, and projected into a relative shift in the FFF profile. Monitor units for all remaining segments were then adapted on-the-fly during delivery. A film insert (3cm spherical target) was positioned 10cm off-centre in the body oval. A liver SBRT plan was created (3x20Gy, 15-beam).
From theory, the average dose scaling for this plan only exceeds 1% for shifts >20mm.
A local Gamma analysis (2%/2mm/2Gy threshold) was performed to quantify dose differences w.r.t. the static reference case. Without table shift, MLC-tracking resulted in a Gamma pass-rate of 99.0%. The 2.5cm table shift resulted in 99.9% with vs. 89.6% without dose scaling, whereas the 5cm table shift resulted in 90.0% vs 60.0%.
The dosimetric performance of MLC-tracking on the Unity MR-linac for motion with very large shifts can be dramatically improved by real-time adaptive dose scaling.
Funding Support, Disclosures, and Conflict of Interest: MF Fast acknowledges funding by the Dutch Research Council (NWO) through project #17515 (BREATHE EASY).