A Comprehensive Characterization of Dosimetry Methods and Beam Quality for In-Vivo Applications of FLASH Radiation Therapy

Abstract

Purpose: A topic at the current forefront of radiation therapy is that of the FLASH effect, which has been shown to both provide tumor control as well as increase sparing of healthy tissue. While the traditional threshold for the FLASH effect to be seen is a mean dose rate (MDR) of 40 Gy/s, the FLASH experiments at Duke University have been able to achieve MDRs ranging from 0.15 Gy/s up to 30 MGy/s using the High Intensity Gamma Source (HIGS) at Triangle Universities Nuclear Laboratory (TUNL). This unique opportunity has allowed previous experiments to delve into the mechanisms behind the FLASH effect using both ex-vivo and in-vivo models. While the majority of prior work has relied on the stability of Gafchromic EBT film for its dose measurement, the latest research calls into question the reliability of this dosimetry form across all dose rates, particularly MDRs of extreme range such as those seen in experiments at Duke. Additionally, the unique dose deposition properties of the HIGS beam have yet to be explored fully through Monte Carlo modeling and clinically replicable beams. The purpose of this work is to advance the methods of dosimetry for FLASH experiments and better understand the beam characteristics that produce this novel effect. This work aims to produce another, near real time, reliable method of dosimetry in comparison to film measurements. It also investigates the consistency and stability of the HIGS beam across various prescribed doses and dose rates, as well as evaluates the accuracy of the HIGS beam in whole brain in-vivo mice irradiations. The work also aims to produce a viable clinically comparable beam to be used in future irradiations as a comparison to clinical outcomes. Methods: The HIGS beam consists of an electron linac which accelerates the electrons to an energy of 35 MeV in single pulses lasting roughly a microsecond. Using combinations of these pulses, we are able to achieve a wide range of dose rates. A 2D translation stage and accurate targeting protocols were used to move between subjects irradiated with the same dose. In order to create a method of near real time dosimetry, Beryllium Oxide Optically Stimulated Luminescent Detectors (OSLDs) were irradiated at various doses. EBT-XD film was also irradiated under the same beam and their resulting dose measurements compared. A similar technique was used to explore the efficacy of a solid-state micro-diamond detector and flashAdapter from PTW. Both the charge readout in single pulse shots as well as charge consistency over multiple pulses were recorded. To test the reliability of the HIGS beam, film analysis was done comparing isodose plots and beam profiles across multiple doses and over multiple irradiations. These analyses were combined with a Monte Carlo model in TOPAS of the HIGS linac to evaluate the beams coverage for whole brain in-vivo mouse irradiations. They were also used to create an experimental electron cutout designed to mimic the size and shape of the HIGS linac for use on clinical accelerators. The characterization of the cutout using EBT-XD film as well as an Eclipse Monte Carlo simulation were used to determine the cutouts ability to mimic to the HIGS beam. In addition to these many dosimetric focused studies, cohorts of C57BL/6J mice were irradiated in two arms. The first contained mice injected with B16F10BrM-Lu tumors and irradiated at dose rates of conventional (~0.15 Gy/s), FLASH (~40 Gy/s) and an ultra-high dose rate termed Super-FLASH (~20 MGy/s). Their corresponding doses were recorded and their weights monitored post irradiation. The second arm contained healthy mice irradiated using an identical setup, whose weights were also monitored. Film used to measure dose to the in-vivo subjects was also analyzed to determine targeting accuracy and develop a system for monitoring the beams coverage of the whole brain irradiations.Results: OSLD read out dose was extremely linear over all doses produced on the HIGS in relation to measured film dose. The discrepancy between the measured dose on film and the measured dose using OSLDs was 25%. The diamond detector exhibited immense variation in charge readout under the HIGS beam despite a predicted linear increase with dose. Film analysis of the HIGS irradiations demonstrated that the beam size, shape and dose distribution remains constant over all doses and is consistent across multiple pulses of the same dose. The electron cutout demonstrated consistency with the HIGS 50% isodose line but failed to replicate the high dose range. The TOPAS Monte Carlo simulation accurately replicated the film profiles seen on the HIGS. It also showed the 80% isodose coverage of whole brain mouse irradiations. Both tumor and healthy mice irradiated under the HIGS beam using the FLASH and Super-FLASH dose rates demonstrated longer survival and more consistent weight than those irradiated using the conventional dose rate. The beam accuracy analysis found that the beam center was below the zygomatic arch midpoint by 2-4 mm on average and 1-2 mm to the right of the midpoint. Conclusion: A near real time dosimetry method was produced for the Duke FLASH experiments using the OSLD reading to predict film doses. The discrepancy between the OSLD dose and the measured film doses is still under investigation. Irradiations of the micro-diamond detector demonstrated its inability to be used for these ultra-high dose rate scenarios. Film analysis showed the stability of the HIGS beam across multiple doses and pulses, an encouraging result for the current in-vivo irradiations. Future work will look at reconciling the OSLD reading and film measurements as well as confirm the HIGS beams positional variance between different doses.