Automated lattice radiation therapy treatment planning personalised to tumour size and shape

Lattice radiation therapy (LRT) alternates regions of high and low doses inside the tumour. Whilst this technique reported positive results in tumour size reduction, optimal lattice parameters are still unknown. We introduce an automated LRT planning method personalised to tumour shape and designed to allow investigation of lattice geometry.

Patients with retroperitoneal sarcoma were considered for inclusion. Automation was performed with the Eclipse Scripting Application Interface (v16, Varian Medical Systems, Palo Alto). By iterating over vertex size (V) and centre-to-centre distance (D), vertices were segmented within the gross tumour volume (GTV) in an alternating square pattern. Iterations stopped when the number of inserted vertices was contained between a prespecified lower and upper bound. Forty sets of lattices were considered, produced by varying V and D in five lower/upper bound pairs. Best-scoring sets were determined with a score favouring the maximization of GTV dose uniformity and heterogeneity whilst minimizing the maximum dose to organs at risk.

Fifty patients with tumour volumes between 150 cm³ and 10,000 cm³ were included. Best-scoring sets were characterised by a low number of vertices (<15). Based on the best-scoring set, the predicted parameters to use for new patients were V = 0.19 (GTV volume)^1/3 and D = 2V, in centimetres. The number of vertices (N) to insert in the GTV can be estimated with N ≤ (24 × 3% GTV volume)/(4πV³).

The automated LRT treatment planning personalised to tumour size allows investigation of lattice geometry over a large range of GTV volumes.