Collision prediction for intracranial stereotactic radiosurgery planning: An easy-to-implement analytical solution

Gantry collision is a concern in linac-based stereotactic radiosurgery (SRS). Without collision screening, the planner may compromise optimal planning, unnecessary re-planning delays can occur, and incomplete treatments may be delivered. To address these concerns, we developed a software for collision prediction based on simple machine measurements. Three types of collision were identified; gantry-couch mount, gantry-couch and gantry-patient. Trigonometric formulas to calculate the distance from each potential point of collision to the gantry rotation axis were generated. For each point, collision occurs when that distance is greater than the gantry head to gantry rotational axis distance. The colliding arc for each point is calculated. A computer code incorporating these formulas was generated. The inputs required are the couch coordinates relative to the isocenter, the patient dimensions, and the presence or absence of a circular SRS collimator. The software outputs the collision-free gantry angles, and for each point, the shortest distance to the gantry or the colliding sector when collision is identified. The software was tested for accuracy on a TrueBEAM® machine equipped with BrainLab® accessories for 80 virtual isocenter-couch angle configurations with and without a circular collimator and a parallelepiped phantom. The software predicted the absence of collision for 19 configurations. The mean absolute error between the measured and predicted gantry angle of collision for the remaining 61 cases was 0.86 (0.01–2.49). This tool accurately predicted collisions for linac-based intracranial SRS and is easy to implement in any radiotherapy facility. La collision avec la tête de l’accélérateur constitue une problématique en radiochirurgie stéréotaxique (SRS). Pour éviter le risque de collision, le dosimétriste peut compromettre la qualité de la dosimétrie, des retards de replanification inutiles peuvent survenir et des traitements incomplets peuvent être administrés. Pour répondre à ces préoccupations, nous avons développé une méthode analytique basée sur de simples mesures pour détecter les collisions. Trois types de collision ont été identifiés : Tête – Table, Tête – Support SRS et Tête – Patient. Des formules trigonométriques pour calculer la distance entre chaque point potentiel de collision et l’axe de rotation de la tête ont été générées. Pour chaque point, une collision se produit lorsque cette distance est supérieure à la distance entre la tête et