AI generates personalized radiopharmaceutical therapy doses in under 23 seconds
Researchers at the University of Massachusetts Amherst have developed an artificial intelligence model that has the potential to improve on the one-size-fits-all dosing approach for radiopharmaceutical therapy (RPT) treatments used in prostate cancer care.
Unlike traditional radiation therapy, RPT is administered via injection. Imaging shows where the radiopharmaceutical accumulates post-injection, but accurately determining how much radiation individual tissues absorb during this process can be time-consuming and difficult.
“The main issue with many cancer treatments is toxicity,” Joyita Dutta, PhD, MS, professor in the Riccio College of Engineering at UMass Amherst, noted in a news release shared Aug. 4. “Whether it’s radiation, chemo or radiopharmaceutical therapy—whatever mechanism kills the cancer cells also hurts healthy tissue.”
Measuring how much radiation tissues have absorbed helps providers understand how individual patients respond to treatment. It also enables them to plan future treatments by giving them the necessary information needed to adjust patient doses. This is where the team’s AI model comes in.
Dubbed DiffuDose, the model generates patient-specific radiation dose maps for RPT. It does this by combining two AI modules—one that creates an initial radiation dose estimate and another that refines it into a high-resolution dose map. It does so this with accuracy comparable to the current gold-standard computational method and in less than 23 seconds per patient. Comparatively, conventional dosimetry calculations can take hours.
In clinical testing, the model outperformed six competing approaches and consistently produced accurate dosimetry across multiple organs, including the kidneys and liver, which are known to be vulnerable to radiation-related toxicity.
“Pixel by pixel in a full image, you could see how the dose was distributed across the body,” said Dutta. “That’s what really helps you personalize the treatment.”
The team has plans to expand the technology through a partnership with the University of Massachusetts Chan Medical School. That collaboration will analyze the utility of combining post-treatment imaging and blood biomarkers to better predict individual patients' response to RPT.
In addition to her background in journalism, Hannah also has patient-facing experience in clinical settings, having spent more than 12 years working as a registered rad tech. She began covering the medical imaging industry for Innovate Healthcare in 2021.