ESTRO 2024 - Abstract Book

S4087

Physics - Inter-fraction motion management and offline adaptive radiotherapy

ESTRO 2024

The PreciseART commissioning process encompassed three main phases: sensitivity analysis, assessment of the dose calculation accuracy and end-to-end testing. In the first two phases the Tomotherapy Cheese phantom was used. Sensitivity analysis [1] included a control plan and plans simulating different error conditions: incorrect couch position, wrong phantom density value, expansion/deformation of the phantom external contour and positioning errors (lateral shifts of ±1, ±2 and ±3 mm). These plans were created in PRECISION 2.0.1.1 treatment planning system, where a cylindrical PTV (76.7 cm 3 ) was delineated at the geometric centre of the Cheese phantom. The irradiation was performed at a Tomotherapy HD unit, positioning the Cheese phantom in the conditions established in the control plan. The planning dose and dose of the day were compared for all plans. To determine the dose of the day accuracy, all of aforementioned plans were recalculated in irradiation conditions and the calculated dose was compared with the reported dose of the day. For plans with lateral shifts, a geometric parameter, Centroid (X,Y,Z), was also evaluated. End-to-End testing was performed with the anthropomorphic Alderson-Rando phantom, for which a head and neck plan was created based on a set of structures that included a molded adenopathy of 20 cm3. First, the phantom was irradiated as per CT images acquisition and then the volume of the adenopathy was reduced to 14 cm 3 . The quality of deformable registration algorithm [2] for all structures was assessed using Dice and Jaccard indices. Dose statistics of all PTV and OAR were calculated for the offline dosimetric assessment taking into account the dose of the day.

Results:

In the sensitivity analysis, the PreciseART was able to correctly identify the error conditions, since the dose difference for the evaluated dosimetric parameters (Dmean, D98%, D2%) exceeded the considered tolerance of 1%. For 1, 2 and 3 mm lateral positioning errors, the Centroid X values agreed with the introduced shifts (0.8, 1.7 and 2.9 mm, respectively).

For the accuracy tests, the mean dose difference for the evaluated parameters was -0.57%±0.77%. The maximum difference was noticed for the case where the external contour was expanded, -2.22%.

For End-to-End testing, Dice value was 0.94±0.03 and Jaccard 0.89±0.06 when considering the initial volumes. For the adenopathy reduction case, the Dice and Jaccard indices reduced from 0.94 and 0.89 to 0.84 and 0.72, respectively. The comparison between the TPS calculation dose and the reported dose of the day showed a mean deviation below 3.0% for the initial volumes. When considering the adenopathy shrinking situation, the target volume coverage decreased 6.0% and the ipsilateral parotid mean dose increased 4.1%.

Conclusion:

In this work a comprehensive commissioning program was followed for offline ART using PreciseART. Patient positioning errors, density and volumes changes were correctly identified. PreciseART was also able to accurately determine the dose of the day and provided dose alerts when significant anatomical changes were verified. Thus, this module can be used for offline ART in the clinical routine together with dose alerts for replanning.

Keywords: head and neck, adaptive radiotherapy, tomotherapy

References: