ESTRO 2024 - Abstract Book

S3626

Physics - Dose prediction, optimisation and applications of photon and electron planning

ESTRO 2024

• predict the GTV/PTV coverages of the new plan, using knowledge from the known 37 plans • rank the known 37 plans from the most similar to the least similar to the new plan. To do that, the tool automatically computes the distance between the 3-uple of the new plan against 3-uples of each of the known plans, and sorts in ascending order.

Results:

A moderate-strong significant correlation between EIV and both PTV and GTV coverages was found, as shown in figure 1, where GTV V47,5Gy and PTV V40Gy are reported, together with linear regression analysis. The Spearman’s Rank Correlation results were respectively -0.59 (GTV) and -0.76 (PTV) (p<0.001). An analogous behavior was observed for the other investigated outputs. When setting up the routine, to improve its prediction capability, GTV and PTV dimensions (in cc) were added as inputs, since an interplay effect with these two parameters was observed. In figure 2 an example picture of the program interface is shown.

The tool was tested for three new patients’ plans, and the results are shown in table 1, in which the comparison between predicted and planned results are compared.

Inputs

Outputs

PTV V40 [%]

PTV Vout40 [cc]

PTV V35 [%]

GTV V47,5 [%]

GTV Vout47,5 [cc]

GTV V50 [%]

EIV [cc]

GTV [cc] PTV [cc]

P1

84,41

111

28,11

90,47

8,95

96,35

88,06

8,81

75,83

Predicted

91,60

9,32

96,8

85,80

11,56

72,70

Planned

P2

10,65

34,33

0,66

99,56

0,00

100

100

0,00

100

Predicted

99,10

0,31

99,9

100

0,00

99,60

Planned

P3

29,9

54,5

1,19

99,01

0,13

99,97

100

0,00

97,29

Predicted

99,80

0,11

100

99,2

0,24

96,80

Planned

Table 1: Comparison between predicted and planned results for three patients