Quantum Amplitude Estimation for Proton Therapy: Samsung Medical Center Preprint of 10 November 2026 Puts a Clinical-Scale Circuit 160 Times Over Its Gate Budget
What Kwangzoo Chung built: a quantum walk that reproduces the Bragg peak
Quentir Medicine Monitor
Evidence-based insights for quantum medicine.
A medical physicist at Samsung Medical Center in Seoul has built a quantum circuit that simulates protons slowing down in tissue and has used it to test quantum amplitude estimation, the quantum method that promises to reach a given precision with far fewer samples than ordinary random sampling. The circuit worked in simulation. At the size a real treatment plan would need, it would cost about one million logical gates per call. Under the most favorable assumptions in the paper, that is 160 times more than a quantum computer could spend per call and still finish ahead of a graphics card.
Kwangzoo Chung, of the hospital's Department of Radiation Oncology and the paper's sole author, posted the work to arXiv on 10 November 2026 and has submitted it to Physics in Medicine & Biology. It has not yet been peer reviewed. For a radiation oncology department, the most useful part follows the circuit: three tests that any claim of quantum speed-up for Monte Carlo dose calculation has to pass, worked out with explicit numbers.
Proton therapy delivers most of its dose at the end of the beam's path, in a narrow region called the Bragg peak, and stops there. That property spares tissue behind the tumor, and it makes the exact range of the beam a safety question. Monte Carlo simulation, which follows millions of individual protons through a model of the patient, is the reference method for checking that range. Its error shrinks only with the square root of the number of simulated protons, so each extra decimal digit of precision costs a hundredfold more computing. Quantum amplitude estimation shrinks the error in proportion to the number of calls, a near-quadratic gain that Ashley Montanaro proved in 2015 for Monte Carlo methods whose output has bounded variance.
Earlier quantum proposals for radiation transport handled photons or neutral particles and carried no energy variable, so they could not represent how a proton loses energy step by step. Chung's circuit adds an energy register. At each step a fresh "coin" qubit, whose rotation depends on the proton's remaining energy, decides whether the proton loses one or two units of energy as it advances. Range straggling, the spread in where individual protons stop, then follows from the circuit itself.
In exact simulation on a classical computer, the circuit produced a Bragg-shaped depth profile that agreed with a classical Monte Carlo run of the same rules to within 0.27 percent. Chung then posed one clinically shaped question to it: the probability that a proton travels beyond a set depth near the distal edge of the beam. He estimated that probability with maximum-likelihood amplitude estimation on reduced instances of 11 and 14 qubits. The error fell almost in proportion to the number of calls, with a fitted exponent of 0.93 against 0.49 for ordinary sampling, and at the largest budget tested it was 4.8 times smaller than sampling at the same number of calls.
No quantum hardware was used and no noise was modeled. The physics is one-dimensional and leaves out multiple Coulomb scattering and nuclear interactions, both of which shape a clinical depth-dose curve. The model's ratio of peak dose to entrance dose cannot exceed 1.69, where measured proton curves reach roughly 3 to 5.