Room: Karl Dean Ballroom A1
Purpose: Describe relative advantages of improved metrics for stereotactic radiosurgery (SRS) plan evaluation and optimization.
Methods: Current metrics used for evaluating SRS plan quality, such as conformity and gradient indices (CI and GI) are impacted by target volume, have no clear relationship to normal tissue toxicity, and provide no straightforward method for calculation of cumulative metrics for multi-target plans. Additionally, GI is dependent on target conformity. Finally, overlapping isodose volumes from separate targets can create nonsensical CI and GI values, not indicative of true plan quality. We define novel illustrative and quantitative metrics for SRS plan evaluation and optimization that improve plan quality assessment and are not subject to the limitations of current indices.
Results: The differential dose volume histogram (dDVH) for total brain minus all target volumes facilitates visual evaluation of the normal brain dose distribution. Scaling this dDVH using equivalent dose in 2Gy fractions (EQD2) assigns emphasis to higher dose regions in normal tissue. For quantitative evaluation, we calculate area-under-curve (AUC) of the normal brain dDVH scaled by EQD2 (α/β=2), and calculate normal brain gEUD (a=4.5). The AUC effectively represents mean EQD2-weighted normal brain dose, while gEUD takes into account the volume effect and provides a normal tissue metric predictive of biological effects. To facilitate visual comparison of SRS plans, we use absolute dose difference maps and subtraction of EQD2-weighted dDVHs. Dose difference maps provide spatial information of variances in both target and OAR dose distributions while differences in EQD2-weighted dDVH offer an enhanced assessment compared to mean dose or cumulative DVH metrics.
Conclusion: Future automation and optimization techniques in SRS planning will drive the need for improved plan comparison metrics. In conjunction with adequate target and OAR dose constraints, techniques presented here accurately quantify plan quality without the limitations of current metrics.