Collision avoidance in computer optimized treatment planning. Academic Article uri icon

Overview

abstract

  • Of major concern in fully automated computerized treatment delivery is the possibility of gantry/couch or gantry/patient collisions. In this work, software has been developed to detect collisions between gantry and couch or patient for both transaxial and noncoplanar treatment fields during the treatment planning process. The code uses the gantry angles, turntable angles, and position of the couch surface relative to the isocenter supplied by the planner for the prescribed radiation fields. In addition, the maximum patient anterior-posterior and lateral separations are entered in order to model the patient outline by a conservative cylindrical ellipse. By accessing a database containing the precise mechanical dimensions of the therapy equipment, 3D analytical geometry is used to test for collisions between gantry/patient and gantry/couch for each treatment field. When collisions are detected, the software inspects the use of an extended distance treatment, by recalculating and testing for collisions, with the couch at a greater distance from the collimator along the direction of the central axis. If a collision is avoided at extended distance, the lateral, longitudinal, and vertical motions of the couch are recorded for entry into the treatment plan, or else a warning message is printed, together with the nearest permissible collision-free gantry angle. Upon inspection, the planner can either elect to use the calculated closest permissible gantry angle or reject the plan. The software verifies that each proposed treatment field is safe, but also that the transition between fields is collision-free. This requires that the sequence of the treatment fields be ordered, preferably into a sequence which minimizes the delivery time compatible with patient safety.(ABSTRACT TRUNCATED AT 250 WORDS)

publication date

  • July 1, 1994

Research

keywords

  • Accident Prevention
  • Radiotherapy Planning, Computer-Assisted

Identity

Scopus Document Identifier

  • 0028093466

PubMed ID

  • 7968836

Additional Document Info

volume

  • 21

issue

  • 7