The Exploration Of FLASH-IMPT Mode:Ridge Filter Design And Its Verification Platform

Flash therapy（FLASH-RT）is a new type of radiation therapy that uses ultra-high dose rates（usually higher than 40 Gy/s）to irradiate the target area.This treatment method can effectively reduce the damage to the normal tissue,and at the same time,the killing effect on the tumor will not be weakened.The successful clinical application of the FLASH therapy is a great innovation to the existing radiation therapy methods.Pre-clinical trials for flash therapy have been carried out extensively in the fields of electronics,photons and protons.Due to the unique dosimetry advantage（Bragg Peak）and intrinsic high dose rate of the protons,it is more suitable for the FLASH research and its clinical application.Additionally,the combination of the two will greatly improve the gain ratio of radiation therapy.Proton FLASH therapy has become one of the hottest areas in the research of radiation therapy.The purpose,method,results and conclusion are as follows:1.At present,the technology of intensity-modulated proton therapy（IMPT）is widely used in clinical treatment,which mainly relies on pencil beam scanning（PBS）technology.However,the time spent by PBS mode in switching proton energy cannot meet the time requirements of flash therapy（generally less than<500 ms）.Passive scattering mode commonly cover the whole target in one shoot.However,so many accessories are added to broaden the beam（in the passive mode）greatly reducing the proton intensity,which limits the possibility of treating a larger target area（generally<5 cm×5 cm）.The combination of range modulators utilized in the passive scattering mode（longitudinal）and pencil beam scanning mode（laterally）can effectively circumvent the above difficulties.2.We have developed a fast beam modulation device（ridge filter）for proton flash therapy under single energy,which can complete IMPT irradiation,greatly shorten the scanning time of the proton beam,while provide a sufficiently high dose rate.First,the design method was verified in the water phantom.The dose distribution formed by the proton beam passing through the ridge filter was compared with the reference dose distribution via three-dimensional（3D）gamma analysis.The 3D gamma passing rate of the static and dynamic ridge filter comparing to the reference is over 95%.The reliability of this design algorithm is confirmed.Based on the self-developed proton planning system（the Geant4 dose calculation module and the optimization module of IPOPT）,the designed static and dynamic ridge filters are applied to actual patient case（H&N case）,and the clinical treatment plans are obtained after dose calculation and optimization.The FLASH related parameters（dose rate,irradiation time）of this radiotherapy plan were analyzed.Applying this design to the H&N and Lung cases,the dose distribution is no big difference from the conventional IMPT plan,and the 3D gamma passing rate exceeds 95%,regardless of static or dynamic ridge filter.For the analysis of parameters related to FLASH therapy,when the proton beam intensity reaches 2.5×10¹¹（H&N）and 8.1×10¹¹（Lung）protons per second,90%of the volume of the PTV can reach the FLASH dose rate specification（>40 Gy/s）.Compared with conventional scanning techniques,adding this ridge filter can shorten the scanning time by 4.9-6.5 times（fraction dose:10 Gy）.The total irradiation time is effectively controlled to less than 1000 ms3.Due to the characteristics of instantaneous high dose（ultra-high dose rate）of FLASH plan,the accurate delivery of FLASH plan is the key to clinical treatment.A slight deviation of position and local dose will cause irreparable harm to patients.Therefore,the plan verification before treatment is particularly important.The fast verification platform based on Monte-Carlo and deep learning was independently developed.First,the Monte-Carlo verification platform was compared with the measured data of known proton facility.The results show that both the depth dose curve and the lateral dose distribution are similar to the measured data（the range consistency is less than 0.85 mm,and the lateral dose difference is less than 2.41%）.Secondly,using deep learning to accelerate Monte-Carlo,its calculation time is shortened from>100min to<1 min.Subsequently,the developed platform combined with Monte-Carlo and deep learning was used for plan verification with multiple patient cases.4.For the case that the treatment plan has not passed the plan verification,it is proposed for the first time to use deep learning platform to create a rapid plan parameter adjustment platform（predict spot weight）.Its test is carried out in a patient case.For the prediction accuracy,the normalized root mean square error（NRMSE）is about 0.41%for data sets with small fluctuation levels.The good accuracy of the machine parameter prediction model is demonstrated.Generally speaking,the design of the ridge filters（static and dynamic）eliminates the need for switching energy for the conventional proton PBS mode,greatly shortening the total irradiation time while maintaining a high dose rate to meet the requirements of FLASH therapy mode.The scanning technology based on this ridge filter is a feasible method for clinical application of proton FLASH therapy.