Fifty-eighth annual meeting of the american association of physicists in medicine
SU-F-T-605: Implementation of the Conformity/Gradient Index (CGI) to Intracranial, Linac-Based Stereotactic Plans to Evaluate Possible Improvements in Treatment Planning
To retrospectively evaluate the planning quality of a sample of intracranial, linac-based, dynamic arc MLC and VMAT stereotactic plans by calculating the effective gradient, Conformity Score (CGIc) and Gradient Score Index (CGIg), and determine if the Conformity/Gradient Index (CGI) may be a useful metric to guide our current treatment planning process.
We evaluated single lesion VMAT, single lesion dynamic arc, and single isocenter, 2-lesion linac-based, intracranial, VMAT treatment plans originally created in Eclipse™. The CGI index was calculated for each lesion using the Target Volume (TV), the Prescription Isodose Volume (PIV), and the 50% Isodose Volume. The effective gradient was obtained directly from the Eclipse “gradient measure.”
From the single lesion cases, target volumes of ≤ ∼2cc reliably achieved an effective gradient of ∼3mm and a CGI > 95. For lesions ≤ ∼20cc, the majority of cases have an effective gradient of <5mm and a CGI > 90-95. For our multilesion data, we determined that lesions separated by <2cm typically had CGI values of 75-85, lesions separated by 2cm-3cm typically had CGI values > 80-85, and lesions separated by >3cm can have CGI over 90.
The CGI metric is a simple tool that can be applied to MLC-based stereotactic planning, requires only three pieces of data, and quickly allows planners to evaluate and determine if a plan is suboptimal. Our data identified several cases that might be improved with additional effort and/or re-planning, but the majority of the treatment plans evaluated fell within a desirable CGI range. Cases deemed suboptimal are currently under investigation and will be re-planned to determine if CGI values can be improved.