The dose algorithm decides the number you are judged on

Small fields in low-density lung break electronic equilibrium, which is exactly where simple algorithms fail. Comparisons across three generations of algorithm in lung SBRT report differences up to 10% against full Monte Carlo — on the same plan, same beams, same MU.

Ranking, roughly:

• Type A (pencil beam) — overestimates dose in lung. Not acceptable for SABR.
• Type B (AAA, collapsed cone) — better, still overestimates in low density.
• Type C (Acuros XB, Monte Carlo) — solves transport; agrees with Monte Carlo in heterogeneous regions.

Why this belongs in a scored challenge. Everyone submits an RTDOSE, and the scorer reads the dose grid it is given. Two planners with identical beams and different algorithms get different scores, and the difference can be larger than the gap between first and fifth place.

What to do about it. Say which algorithm and which grid size you used, in the write-up and on the submission. It costs a line and it is the difference between a comparable result and a number.

Grid size is part of the same problem. A near-maximum metric like D0.03cc is grid-dependent by construction — a 2.5 mm grid and a 1 mm grid disagree on it. Calculate the final dose on a fine grid before reading the metric you will be judged on.