The proton beam is like a sniper's bullet: it delivers most of its energy at a precisely set point, barely touching tissues on the way in. But over 4–6 weeks of head and neck tumor treatment, the target shifts: the patient loses weight, the tumor shrinks, posture changes. The sniper aimed at one spot, but the target has already moved.
The usual approach is to do a weekly control computed tomography scan and recalculate the plan. This takes about a week and adds radiation dose from the scan itself. The research group proposed a different path: build a 'digital twin' of the patient. Take the initial scan of the person and a scan of another patient who has already undergone part of the treatment. Using two successive image registrations, the program figures out how the 'donor's' anatomy changed and transfers those changes to the new patient — resulting in a predicted scan for the day of treatment.
On 88 patients with three control CT scans each, the predicted scans matched the actual anatomy of the treatment day noticeably better than the initial plan. Organ shape agreement improved by a fifth, and the error in CT numbers dropped by nearly a quarter. The biggest gain was for those whose bodies changed the most. For patients with stable anatomy, the prediction was almost identical to the initial scan. Thus, the digital twin prepares the sniper in advance for where the target will move.
The Bragg peak was discovered in 1903 by William Bragg when he irradiated thin layers of mica with alpha particles. Today, this effect is the basis of proton therapy, allowing tumors to be irradiated while barely affecting the skin at the entry point.
In the TV series 'The Expanse', medical pods scan the body and instantly adjust treatment to the current state — the digital twin takes a first step toward such a future.