ESTRO 2024 - Abstract Book

S4916

Physics - Quality assurance and auditing

ESTRO 2024

UCoMX [5-7]. A complexity metrics signature was created to describe four aspects of the plans: the presence of small fields, the total number of MUs delivered, the irregularity of the Beam Eye Views (BEVs), and the amount of leaf transmission. Adopting the approach proposed in [8], a Plan Complexity Score (PCS) was calculated for each plan. It was computed as a linear combination of the subset of principal components (PCs) obtained through Principal Component Analysis (PCA) that allowed explaining the 90% of the total variance. Based on the values of the PCS, the trend of the delivery accuracy (PR γ %(3%G, 2 mm)) and clinical acceptability (PQM%) was investigated. software developed at our institution

Results:

The signature included five metrics: MUs delivered per cGy (MUcGy), Edge Metric (EM), namely the ratio between the BEV’s perimeter in the cross-leaf direction and its area, Average Leaf Gap (ALG), namely the average distance between pairs of opposing leaves, BEV-Jaw Area Ratio (BJAR), which represents a rough measure of the relative weight of leaf transmission dose on the total dose, and Tongue-and-groove, which quantifies the leaves interdigitation. Figure 1 shows that approximately 90% of the variance is explained by the first three PCs (i.e. 51%, 23% and 16%, respectively) used in the final PCS. The first PC is dominated by EM, ALG and MUcGy, principally describing the irregularity of the BEVs and the effect of small fields. The second component is dominated by TG and this shows that the effects of the two irregularity measures in the signature (i.e. cross-leaf with EM vs along-leaf with TG) are disentangled by the PCA and lead to different relative contributions to the overall PCS. The third component is dominated by the BJAR.

Figure 2 shows the average trend of the delivery accuracy and clinical acceptability for increasing values of PCS. When PCS increases, a net decrease of the deliverability in favor of an increased clinical acceptability is observed. Therefore, upon extensive clinical validation, the PCS herein proposed could provide simoultaneous information on clinical acceptability, deliverability and complexity, resulting in a quantitative measure of the overall plan quality. In practice, this would allow defining an optimal region where both high clinical acceptability and delivery accuracy are achieved while keeping the complexity level as low as possible.