ESTRO 2024 - Abstract Book

S3440

Physics - Dose calculation algorithms

ESTRO 2024

conventional radiotherapy planning. Since Cobalt energy has similar HU-to-relative_electron_density (RED) curve to 6MV’s, the RED tables (all of which have included Onyx’s RED) that are used in our institution’s other treatment planning system are entered into GammaPlan. Once the electron density has been properly defined, GammaPlan can report dose statistics (Min, Max, Mean) and Dose Volume Histogram (DVH) calculated from both TMR10 and convolution algorithms of the clinical plan simultaneously (BOT is kept the same). The dose distribution from each calculation can also be individually exported in DICOM format (RTDOSE) for comparison at other systems. Besides Tumor volume (TV), Min, Max, and Mean doses of tumor and OARs (organs at risk) reported from Volume under Measurements , we also recorded from DVH the following metrics: TV Rx (TV receiving Rx dose), PIV (prescription isodose volume), and PIV 1/2 (half-prescription isodose volume) in order to calculate Coverage (TV Rx /TV), Selectivity (TV Rx /PIV), and GI (Gradient Index = PIV 1/2 /PIV) for both TMR10 and convolution calculations. This process is repeated for the pre- and post-Onyx CT for each patient. Consistently, compared to TMR10, convolution calculation results in lower Min, Max, Mean, TV Rx , PIV, PIV 1/2 , and Coverage, but higher Selectivity (mainly due to lower PIV) and GI. The differences in (Mean, Coverage , PIV, Selectivity , GI) range from (-5.1 – -9.6, -1.8 – -9.4, -9.3 – -23.2, 7.8 – 17.9 , 3.1 – 11.8)% in pre-Onyx comparisons to (-5.4 – -10.6, - 0.3 – -11.3 , -8.2 – -26.0, 7.4 – 19.8 , 0.9 – 14.9)% in post-Onyx comparisons. The convolution algorithm calculation has a loss of coverage up to 10% or more. Individually, the deviations of post-Onyx convolution from TMR10 is slightly higher than the deviation of pre-Onyx convolution from TMR10, by 0.5% for patients with smaller Onyx volumes and ~3% with larger Onyx volumes. The ranges of Onyx volumes are (0.076 – 1.911) cc. Before applying the clinical RED curve, we verified that, when the RED table was set to all water, convolution indeed calculated the same results as TMR10 (< 0.5% difference). For OARs, convolution algorithm consistently calculates lower (Min, Max, Mean) doses than TMR10 with a maximum difference of 0.9 Gy for pre-Onyx and 1.1 Gy for post-Onyx. Results:

Conclusion:

There are significant differences between convolution and TMR10 calculated doses of the same plan (same shot locations/patterns/weights and BOTs). The differences of convolution calculations performed on pre- and post-Onyx CT are much less and become insignificant for small Onyx volumes. Past clinical experiences and prescriptions have been based on TMR10 calculated results. Therefore, switching to convolution algorithm may necessitate a reduction in prescription dose to match past experiences and will require more studies to confirm.

Keywords: Gamma Knife, Icon, TMR10

References:

1. Rojas-Villabona A, Kitchen N, Paddick I. Investigation of dosimetric differences between the TMR 10 and convolution algorithm for Gamma Knife stereotactic radiosurgery. J Appl Clin Med Phys. 2016 Nov 8;17(6):217-229. doi: 10.1120/jacmp.v17i6.6347. PMID: 27929495; PMCID: PMC5690517. 2. Xu AY, Bhatnagar J, Bednarz G, Niranjan A, Kondziolka D, Flickinger J, Lunsford LD, Huq MS. Gamma Knife radiosurgery with CT image-based dose calculation. J Appl Clin Med Phys. 2015 Nov 8;16(6):119–129. doi: 10.1120/jacmp.v16i6.5530. PMID: 26699563; PMCID: PMC5691031.