Dosimetric assessment of conventional and advanced algorithms in clinical stereotactic radiotherapy

To systematically compare the dosimetric performance of conventional (Ray Tracing, AAA) and advanced (Monte Carlo, Acuros XB) dose calculation algorithms across homogeneous and heterogeneous tissues in stereotactic radiotherapy (SRT) and stereotactic body radiotherapy (SBRT).

A retrospective analysis of 125 SRT cases (brain: 50, lung: 20, liver: 20, spine: 35) was conducted using CyberKnife and Varian systems. Plans were originally created using Type B (Anisotropic Analytical Algorithm [AAA] and Ray Tracing) algorithms and were subsequently recalculated using Type C (Acuros XB and Monte Carlo) algorithms, while maintaining identical beam geometry and monitor units. Dosimetric parameters (D_95%, D_mean, D_max, CI, HI, GI) were evaluated. Validation included point dose measurements with Cheese Phantom and gamma index analysis using the PTW 1600 SRS Phantom.

In lung cases, Type B algorithms overestimated D_95% by 14% compared to Monte Carlo, which reduced D_mean by 13.7% and CI by 25.8%. In liver, Acuros XB lowered D_mean by 21.4% with a 0.8% CI reduction. For spine, Monte Carlo reduced D_95% by 3.4%, with a 1.1% drop in D_mean and stable CI. Brain cases showed minimal differences, with Monte Carlo increasing CI by 2.5% (1.19 vs. 1.16). Gamma pass rates exceeded 98% for Monte Carlo and Acuros XB, surpassing Ray Tracing and AAA (≤96%).

Advanced algorithms demonstrated superior dose accuracy, homogeneity, and organs at risk (OAR) sparing in heterogeneous anatomical regions. Despite higher computational requirements, their clinical implementation is justified for SRT/SBRT planning. This study supports a site-specific approach, advocating for advanced algorithm use in anatomically complex scenarios.