Normal tissue often gets caught in the crossfire during radiation therapy. Damage is caused by the high-energy beams of radiation used to kill tumor tissue–particularly when the patient’s breathing causes the tumor to shift.
To better track a tumor’s position in real time and adjust the radiation accordingly, researchers at the University of Alberta in Canada have combined a linear accelerator with a magnetic resonance imager. Today in Anaheim, CA, at the annual meeting of the American Association of Physicists in Medicine, researchers will present evidence that a device that combines these technologies can accurately track and irradiate a moving target.
Radiation therapy uses high-energy x-rays from a medical linear accelerator to damage tumor tissue and treat nearly every type of cancer. In the United States, half of all patients with cancer receive this form of treatment, which typically requires 10 to 15 sessions lasting from about 15 to 30 minutes each. In order to make sure the entire tumor is irradiated, doctors have to irradiate a margin of healthy tissue around it, which leads to side effects including nausea, pain, and skin-tissue damage. In between sessions, the healthy tissue regenerates, but the tumor does not. One way to minimize the side effects is to lower the radiation dose and increase the number of sessions, sometimes to as many as 35.
“We would like to decrease the margins and increase the radiation dose, in order to control the tumor better without side effects,” says Gino Fallone, director of the medical physics division at the University of Alberta department of oncology.
Another challenge is posed by tumor movement during treatment. Tumors in the lungs and the prostate especially may move by about two centimeters during treatment. Current radiotherapy deals with this challenge by combining the radiation source with a computed tomography (CT) scan. This helps doctors reduce damage to healthy tissue, but CT scans are not very good at showing soft tumor tissue, and they are too slow to track tumor movement in real time. Fallone’s group has turned to magnetic resonance imaging (MRI), which provides crisp pictures of soft tissues such as tumors, in the hopes of doing better.
Until now, it hasn’t been possible to use MRI to guide radiotherapy. This is because MRI machines and the linear accelerators that supply high-energy x-rays for radiotherapy interfere with each other. MRI uses a strong magnet and pulses of radio-frequency waves to excite and read a signal from protons in the water molecules inside soft tissues in the body. Medical linear accelerators also use radio-frequency pulses, in their case in order to accelerate electrons through a waveguide toward a metal target. When the electrons hit the target, high-energy x-rays come out the other side; these x-rays are then aimed at tumor tissue. If these two machines are in the same room, the magnetic field from the MRI interferes with the waveguide, preventing the electrons from being accelerated, and the radio-frequency pulses from the linear accelerator interfere with the imager’s magnetic field, degrading picture quality.
To combine the technologies, the Alberta researchers had to reengineer both components. “The whole machine is designed differently,” says Fallone. Special shielding is employed. And instead of using a high-strength magnetic field generated by superconducting-wire coils, as in clinical MRI, the machine uses a weak permanent magnet. The weak magnet interferes much less with the accelerator and is smaller and less expensive to operate. This December, Fallone’s group published the results of imaging studies that showed it was possible to generate MRI images while running the linear accelerator without interference.
The weak magnet imposes a different challenge, however: the image quality is much lower. So researchers at Stanford University are working on computational methods for getting the necessary information from these lower-resolution images. “Diagnostic MRI requires a very high image quality, but for radiotherapy you don’t need to see the tumor in exquisite detail,” says Amit Sawant, an instructor in radiation oncology at the Stanford School of Medicine. “You can afford to lose [image] signal, and still get enough information to know when the tumor is moving.” What’s important to see during radiotherapy, says Fallone, are the edges of the tumor.
Fallone and Sawant will present initial results of image-tracking studies done with the prototype combined device at the conference in Anaheim. Sawant’s group will describe imaging software that allows the machine to acquire five two-dimensional MRI images per second–much faster than conventional MRI. The Stanford researchers increased the imaging speed by decreasing the imaging area and using a technique called compressive sensing. When images are stored, about 90 percent of the data is thrown out; using compressive sensing, it’s possible to acquire only the most important 10 percent of the image data in the first place.
Fallone will present results demonstrating that such real-time guidance can be used to redirect the prototype device’s x-ray beam. “So far, only CT has been available for image guidance,” says Bhadrasain Vikram, chief of the clinical radiation oncology branch of the National Cancer Institute’s Radiation Research Program. “It’s exciting that [MRI] is becoming available to start asking whether it can provide more accurate information.” Better guidance for radiotherapy, says Vikram, might speed up the treatments or even “cure some cancers you can’t cure today.”
But before the system can be tested on patients, the researchers caution that the image-acquisition process needs to be sped up even more, so that it’s possible to make 3-D images. The device will also need to be tested on animals. Fallone estimates that human tests are at least five years away.
This new data poisoning tool lets artists fight back against generative AI
The tool, called Nightshade, messes up training data in ways that could cause serious damage to image-generating AI models.
The Biggest Questions: What is death?
New neuroscience is challenging our understanding of the dying process—bringing opportunities for the living.
Rogue superintelligence and merging with machines: Inside the mind of OpenAI’s chief scientist
An exclusive conversation with Ilya Sutskever on his fears for the future of AI and why they’ve made him change the focus of his life’s work.
How to fix the internet
If we want online discourse to improve, we need to move beyond the big platforms.
Get the latest updates from
MIT Technology Review
Discover special offers, top stories, upcoming events, and more.