Complexity metrics, and what they actually predict
MU per Gy is the crude measure. MCS is the one with a published correlation to QA failure — and the correlation is only moderate.
Built in·updated 2026-08-03
Modulation is not free: at some point a plan stops being deliverable as calculated. The measures in use, roughly in order of how much they tell you:
- Total MU and MU per cGy — crude, but available from any plan and comparable within a prescription.
- Number of control points / segments.
- Modulation Complexity Score (MCS) — combines aperture area variability, leaf sequence variability and MU weight per segment. The arc version is MCSv.
What the evidence supports. Most complexity metrics correlate weakly with gamma passing rate. MCSv is the exception, at moderate correlation — Spearman's ρ around 0.50–0.53 on two TrueBeam platforms. That is enough to flag plans worth measuring and not enough to replace measuring them.
What this means for a challenge. A deliverability score built on total MU is measuring the crudest available signal. It is still worth having, because the alternative is a leaderboard where the winning plan is one nobody could deliver — but it should be read as "this plan is more modulated than that one", not as "this plan will fail QA".
Practical. Report MU per prescribed Gy rather than raw MU, so a 30 Gy SABR plan and a 70 Gy head and neck plan are on the same axis at all.
Where the numbers come from
- A new metric for assessing IMRT modulation complexity and plan deliverability
Med Phys · 2010
Original MCS: aperture area variability × leaf sequence variability, MU-weighted per segment
- Quantified VMAT plan complexity in relation to measurement-based quality assurance results
J Appl Clin Med Phys · 2020
Most metrics correlate weakly with gamma passing rate; MCSv moderate, ρ = 0.502 and 0.528