(Sunday, 7/12/2020) [Eastern Time (GMT-4)]
Room: AAPM ePoster Library
Purpose: non-coplanar beams are used in radiotherapy treatment planning, dosimetric advantages become available that lead to therapeutic advantages for the patient. However, adding an additional piece of moving equipment or angle of rotation can lead to an increase risk of collisions between the gantry and the patient. Accurate collision prediction models are necessary for patient safety, yet, most models are non-patient specific and leave out parts of the body that are potential risks for collisions. Therefore, the current study proposes to use a deph camera to obtain a patient-specific full-body model to be used in an in-house developed collision script.
Methods: The Azure Kinect is a motion tracking camera capable of providing depth definition through an infrared signal to the surface of the patient’s anatomy. Data from the camera can be extracted and used to create a surface model of the patient. This surface model will then be fused to the external provided by CT Sim to create an accurate representation of the patient. A Rando Phantom will be used to simulate the patient and act as a known surface.
Results: The surface model obtained from the Kinect will be compare to a known surface to validate the Kinect’s accuracy. The known surface will be generated from creation of an external similar to one on a treatment planning CT. The point clouds of the two surfaces will be compared and the standard deviation between points calculated.
Conclusion: Accurate anticipation of collisions will increase patient safety while allowing for innovative care through non-coplanar treatments. The addition of the Kinect as a surface imager during the simulation phase of a patient’s treatment creates a patient-specific representation that is unavailable from a standard CT simulation of the target site.
Keywords
Not Applicable / None Entered.
Taxonomy
Not Applicable / None Entered.
Contact Email
