Leveraging AI to Personalize Spatiotemporal Dosimetry

Caroline Chung, MD, MSc, FRCPC, CIP, noted that while the targeting is personalized, radiotherapy has previously been administered with a "one-size-fits-all" approach: standardized doses delivered on fixed schedules, much like a universal drug prescription. However, considering the growing prevalence of artificial intelligence (AI), this rigid paradigm is shifting toward a more fluid and biological model of care.

In this interview from the 2026 ASCO Gastrointestinal Cancers Symposium, Chung, vice president, chief data and analytics officer, and co-director of the Institute for Data Science in Oncology at MD Anderson Cancer Center, explored how novel AI tools are redefining the field’s understanding of energy deposition and its biological effects in tissue. Chung highlighted the opportunity to transition away from simply treating visible "lumps" toward analyzing the spatial and temporal features of a tumor. By using AI, clinicians can now identify sub-visual characteristics within scans and explore "spatiotemporal dosimetry" to manipulate radiation delivery to protect the normal tissues, including the patient’s immune system, while maximizing the therapeutic impact on the cancer.

Transcript:

Radiotherapy, up until this point, has been largely delivered in a very uniform [manner]. It’s like saying that if you were to take a drug equivalent, you’re saying you can only deliver 200 mg of this dose—no matter who it is, no matter where it is—daily, and that’s the schedule to deliver radiation. That’s a very limited way of delivering energy, which is ultimately what we’re doing when we're delivering radiation; we’re depositing energy into tissue. We’re now starting to explore the entire array of delivering different doses each day, perhaps. [We’re] looking at different ways of defining the targets that we’re treating, not just treating the lump of tumor that we visibly see, but characterizing those tissues based on underlying AI-discovered features within the scan, and looking at the images more from a spatial perspective. These are ways that we can explore spatial, temporal, and dosimetric variations in radiation. We were just discussing sparing certain parts of the tissue intentionally to preserve immune function and immune response because we are starting to discover that the immune response is an incredible part of cancer therapeutic response. Can we manipulate the way that we deliver the radiation spatially, but also temporally, to spare certain regions from certain doses? There are many ways we can explore the drug discovery or clinical trial space, going across the comprehensive range of how radiation can be delivered for cancers. Thinking about the human—the entire human—and their own immune status is an exciting space for us to be exploring. It’s opened a whole world of research.