(Sunday, 7/12/2020) [Eastern Time (GMT-4)]
Room: AAPM ePoster Library
Purpose:
The Conformity/ Gradient Index (CGI) is a simple metric used to evaluate single target SRS plan quality. It is characterized by three data points: the sum of target volumes (TV), prescription isodose volume and 50% isodose volume. The purpose of this work was to study if a CGI prediction model can be useful in evaluating Single Isocenter Multiple Target (SIMT) SRS plans.
Methods:
Data for 477 SIMT clinical plans with up to 30 targets and total volumes ranging from 0.4 to 98.4cc was extracted from the treatment planning system and the corresponding CGI scores were calculated. The correlation of CGI with target volume and number of targets was studied using bivariate linear regression methods. The data was split into training (60%), validation (20%) and test (20%) sets and was randomly shuffled within the training and validation sets to build the prediction model (average of the regression coefficients from 10 reshufflings), which was finally tested against the test set.
Results:
CGI values ranged from 22.5 (23 lesions, 42cc volume) to 99.2 (2 lesions, 1.1cc volume), with an average score of 78.6 ± 14.2. The bivariate regression model showed the CGI linearly decreased with increasing number of targets and total target volume (p-value = 0, a = 0.05, R²=0.90). The average percent difference from the test dataset’s calculated CGI to the model predicted CGI was 1.31 ± 5.28.
Conclusion:
This preliminary work shows the CGI’s applicability to SIMT plan evaluation. Suboptimal plans can be identified if the calculated CGI score lies below the predicted value. Future studies will study its correlation with target shape and distance to isocenter to obtain a more accurate prediction model.
Keywords
Not Applicable / None Entered.
Taxonomy
Not Applicable / None Entered.
Contact Email
