An Ionizing Radiation Acoustics Imaging (iRAI) Approach for guided Flash Radiotherapy - SUMMARY
An emerging radiotherapy (RT) modality that utilizes ultra-high dose rate, known as FLASH-RT, has
demonstrated unprecedented ability for improving RT therapeutic ratio in preclinical studies and early clinical
cases. Because of lack of appropriate image-guidance technologies, these studies have been limited to superficial
irradiations and simplistic cases where monitoring of delivered dose is permissible using existing methods. This
severely handicaps the prospects of FLASH-RT and largely limits its promising impact for deep seated tumors,
which constitute most of RT cancer cases. It is widely recognized that currently used dosimetry technologies fall
short of providing the necessary guidance to deliver FLASH-RT in a practical clinical setting without exposing
the patient to tremendous risks that go far beyond the traditional RT delivery. Undoubtedly, there is an unmet
need to develop in vivo image-guidance techniques to safeguard FLASH-RT accurate delivery. We hold that these
challenges can be resolved by refining the emerging technology of ionizing radiation-induced acoustic imaging
(iRAI), which can be intrinsically paired with FLASH-RT delivery systems. iRAI is based on the known
thermoacoustic phenomenon in radiation physics, where acoustic waves are generated from thermoelastic
expansion of a substance following absorption of penetrating pulsated high energy radiation. Building upon our
multi-institutional multidisciplinary team with expertise in ultrasound (US) imaging, RT physics, data analytics,
and our promising preliminary results, we hypothesize that: (1) a dual-modality imaging system comprised of
iRAI and US (iRAI-US) can simultaneously image both tissue morphology and 3D dose deposition during
FLASH-RT delivery with high spatio-temporal resolutions; and (2) machine learning based reconstruction and
anomaly detection can effectively improve imaging quality and mitigate errors, respectively, for clinical
translation. Therefore, in this project we aim to exploit the technological potentials of iRAI-US and machine
learning for developing an image-guidance platform for effective and safe FLASH-RT delivery. We will
demonstrate its efficacy with electron and proton beams using computer simulations (in silico), tissue mimicking
phantoms, and relevant preclinical in vivo models. Specifically, we will (1) develop and test a dual-mode imaging
system for 3D radiation-acoustics dosimetry and US imaging for FLASH-RT; (2) evaluate the in vivo
performance of iRAI-US dual imaging during electron and proton FLASH-RT deliveries; and (3) adapt and
improve iRAI volumetric representation, temporal resolution and error detection for FLASH-RT using deep
machine learning algorithms (DeepRAI) towards effective clinical implementation.
Impact: Our proposed image-guided FLASH-RT, once validated, will offer a practical, robust, cost-effective,
and unique system for safeguarding FLASH-RT delivery. These advancements will address the current
challenges impeding the clinical translation of FLASH-RT and enable achieving its promise of limiting
radiotherapy toxicity to normal tissues and thereby improving cancer patient care and quality of life.
