Robustness and Decision Value of Functional Liver Volume–Based Dose Metrics In MR-Guided Adaptive Liver SBRT
Abstract
Purpose
MR-guided adaptive liver stereotactic body radiotherapy (SBRT) enables daily visualization of liver anatomy and facilitates functional liver–guided planning through online plan adaptation. Functional liver volume (FLV)–based dose metrics, including V5, V15, and mean FLV dose, have been proposed to supplement conventional whole-liver constraints. However, their robustness and reliability under uncertainty in FLV definition remain unclear. This study evaluates the robustness and decision value of these selected metrics.
Methods
Five patients treated with MR-guided adaptive liver SBRT were retrospectively analyzed. For each, a reference FLV was defined using imaging-derived functional surrogates. To account for definition uncertainty, multiple FLV variants were generated by systematically varying segmentation thresholds and image processing parameters. Metrics including V5(FLV), V15(FLV), and mean FLV dose were calculated. Robustness was quantified using coefficients of variation, while stability of plan ranking and plan preference changes were rigorously assessed.
Results
FLV-based dose metrics demonstrated heterogeneous sensitivity to definition uncertainty, resulting in variable plan ranking stability. Mean FLV dose exhibited substantial variability across variants and was associated with frequent changes in plan preference. In contrast, V5(FLV) and V15(FLV) showed greater robustness and preserved plan ranking consistency. These volume-based metrics consistently supported FLV-sparing strategies without altering clinical plan preference under perturbations, offering more dependable results than mean dose metrics.
Conclusion
In MR-guided adaptive liver SBRT, FLV-based dose metrics differ markedly in robustness and decision stability. Volume-based metrics such as V5(FLV) and V15(FLV) demonstrate superior robustness compared with mean FLV dose, providing more reliable guidance for adaptive planning decisions. These findings underscore the necessity of conducting robustness evaluation when selecting FLV-based dose metrics for clinical implementation in adaptive SBRT workflows to ensure treatment decision stability.