Skip to main content
Medical physics

Medical physics

Best in physics: clinical prompt gamma measurements and FLASH dosimetry

22 Jul 2020 Tami Freeman
Prompt gamma team
Best in Physics presenter Joost Verburg (back row, next to the detector) and colleagues with the prototype prompt gamma detection system on its positioning robot. (Courtesy: Joost Verburg)

The “Best-in-Physics” poster session at the 2020 Joint AAPM|COMP Virtual Meeting highlights the top five abstracts in the imaging, therapy and multi-disciplinary science tracks. Our first pick of this year’s top scoring studies examined multi-target tracking during radiotherapy and carotid plaque evaluation using quantitative ultrasound. In our second look at the 2020 winning studies, we report on the first ever prompt gamma ray spectroscopy in a patient, plus an investigation into the use of ionizing radiation acoustic imaging for in vivo FLASH dosimetry.

First prompt gamma ray spectra measured during proton therapy

Proton therapy offers dosimetric advantages over photon-based radiotherapy, such as reducing integral dose by factor of two or three, while delivering the same dose to the target. It should also be possible to use the sharp distal edge of the Bragg peak to create a highly conformal high-dose area around the target. “However, in practise, we can’t do that yet as we don’t know exactly where the protons stop,” explained Joost Verburg, from Massachusetts General Hospital and Harvard Medical School.

To address this problem, Verburg and colleagues are developing prompt gamma-ray spectroscopy for in vivo verification of proton range. Prompt gamma rays are generated when the therapeutic proton beam interacts with atomic nuclei within the patient. The idea is to measure the range of every proton pencil beam, during treatment delivery, by acquiring energy- and time-resolved prompt gamma-ray spectra. These spectra are compared with a nuclear reaction model to determine the range deviation of the pencil-beam compared with the calculated treatment plan.

Verburg described his team’s prototype system and presented the first measurements of prompt gamma-ray spectra obtained during a patient treatment.

The MGH/Harvard team has created a full-scale prototype prompt gamma-ray spectroscopy system, comprising a collimated array of fast scintillation detectors mounted on a 7-axis robotic positioning system for patient alignment. Initial tests in phantoms revealed a typical measured range error of less than 1 mm. “This shows that in a situation where we actually know the materials and know where the protons end up, we are measuring exactly what we are supposed to,” Verburg noted.

The first patient recruited in the team’s clinical study received proton therapy for a complex base-of-skull meningioma. The researchers measured prompt gamma spectra of one treatment field once per week for five weeks. Results were consistent from week-to-week, with a systematic range deviation between the delivered proton range and the treatment plan.

The mean range error was just 1–2 mm, but Verburg noted that not all pencil beams stopped exactly where they were planned. They observed a spread of around 3 mm sigma and a maximum range error of about 8 mm. Compared with traditional range margins used for proton therapy (3.5% + 1 mm, or 6.6 mm for this case), these errors fit within traditional margins.

“We successfully performed first ever prompt gamma ray spectroscopy in a patient,” Verburg concluded. “The range errors that we measured were mostly consistent between fractions and within traditional range margins that are applied in proton therapy. The test in phantoms also showed that our prompt gamma ray spectroscopy system is capable of aiming protons very precisely with a range position of around 1 mm. This shows great potential – if we measure and fine tune the range of the proton beams in the patient, we can reduce the need for margins and design better proton treatments in future.”

Ionizing radiation acoustic imaging offers in vivo FLASH dosimetry

FLASH radiotherapy is an emerging treatment modality that uses ultrahigh dose rates (above 40 Gy/s) to spare normal tissues and improve the therapeutic ratio compared with conventional radiotherapy. The instantaneous delivery of higher dose rates, however, increases the need for reliable beam localization and dose monitoring tools, particularly when dealing with deep seated tumours, where current dosimeters can’t provide an adequate reading for safe delivery.

Noora Ba Sunbul from the University of Michigan is investigating the use of ionizing radiation acoustic imaging (iRAI) for in vivo FLASH dosimetry. “The main aim of this work is to fully develop a comprehensive simulation workflow to test the feasibility of iRAI as a reliable real-time dosimetry tool for FLASH radiotherapy,” she explained.

When a pulsed beam of therapeutic radiation strikes tissue it causes localized dose deposition and thermal expansion, which generates acoustic waves. iRAI works by detecting these waves with an ultrasound transducer and using them to construct dose-related images in real time.

To test their approach, Ba Sunbul and colleagues used a modified linac to deliver a 6 MeV electron beam in FLASH mode to a gelatin phantom. The field was collimated to 1×1 cm and an ideal transducer placed 10 cm from beam centre. They also simulated the detection of induced acoustic waves following FLASH using Monte Carlo and k-Wave. This involved simulating the full 3D dose distribution in the phantom, using this to define the initial pressure source, modelling the acoustic wave propagation and then reconstructing an image.

“Comparing dose profiles at different depths with our film-measured results showed acceptable agreement between measured and simulated results, with less than 6% error at depths of less than 2 cm,” said Ba Sunbul.

The team also used the instantaneous pressure signal to define the edges of the beam, determined by the change in this pressure signal between the phantom entrance and exit. iRAI could identify the central beam edges within about 4% of the film measurement.

The linac pulse duration and repetition rate, which define the dose rate, were seen to be inversely proportional to the instantaneous pressure signal amplitude, as well as the dose. Ba Sunbul explained that the pulse duration affects both the temporal and spatial resolution of the 2D reconstructed iRAI images, with longer pulses creating lower-resolution images. She noted that this negatively affects the beam localization ability of iRAI.

“We have developed a full simulation workflow for testing iRAI applicability in FLASH radiotherapy using ideal point-source ultrasound transducers,” Ba Sunbul concluded. The next step, she added, will be to simulate iRAI image reconstruction to mimic experimental set-ups, noting that iRAI’s ability to detect beam edges has already been verified experimentally using conventional radiotherapy in rabbit measurements and in a liver phantom.

Copyright © 2024 by IOP Publishing Ltd and individual contributors