Dose fall-off index (DFI): An organ-centric metric for spatial dose gradient evaluation in head and neck IMRT and VMAT

Achieving sharp dose gradients is essential in head and neck radiotherapy to ensure tumor control while sparing nearby organs at risk (OARs). Conventional dose-volume histograms (DVHs) summarize dose but do not show spatial fall-off at OAR boundaries. To address this, we evaluated the Dose Fall-off Index (DFI), an OAR-centric metric that normalizes high-dose subvolume to organ size. To provide clinical context, DFI was examined together with normal tissue complication probability (NTCP) modeling and risk classification.

A total of 160 treatment plans were retrospectively analyzed, including VMAT ( n = 82) and IMRT ( n = 78). DFI was calculated for the parotid glands, larynx, oral cavity, and pharyngeal constrictor muscles (PCM). Correlations with irradiated volume and mean dose were assessed, and subgroup comparisons were made by diagnosis, postoperative status, SIB use, and age. Statistical tests included t-tests, ANO V A, and Bonferroni corrections.

DFI showed an inverse relationship with irradiated volume and a moderate association with mean dose across OARs. Despite comparable mean doses, VMAT achieved significantly higher parotid DFI than IMRT ( P < 0.001), consistent with improved angular dose sampling. Lower DFI values were observed in laryngeal tumors, postoperative cases, and SIB plans, while the weak negative association with age was not clinically meaningful. VMAT plans more frequently fell into the low-risk category on NTCP-based stratification.

DFI is a simple, organ-focused metric that complements DVHs by describing dose gradient steepness. When used alongside NTCP modeling, DFI provides a complementary framework for interpreting spatial dose gradients within the context of established biological risk estimates. Further studies incorporating clinical toxicity outcomes are required to determine its clinical relevance and potential association with treatment-related toxicity outcomes.