Tumor position determination

A method of determining a position of a tumor in a subject, the method comprising:

• providing an initial image of a tumor in a subject and an initial two-dimensional (2D) or three-dimensional (3D) representation of a surface of the subject;
• providing a subsequent 2D or 3D representation of a surface of the subject, wherein the subsequent 2D or 3D representation is taken at a subsequent point in time than the initial image and the initial 2D or 3D representation;
• and
• determining a subsequent position of the tumor in the subject based on the subsequent 2D or 3D representation, the initial 2D or 3D representation, the initial image and a computer-implemented tumor position determining model trained based on a plurality of sets of image data, each set of image data comprises an initial training image of a tumor in a subject, an initial training 2D or 3D representation of a surface of the subject, a subsequent training image of the tumor in the subject and a subsequent training 2D or 3D representation of a surface of the subject, wherein
• the subsequent training image and the subsequent training 2D or 3D representation are taken at a subsequent point in time than the initial training image and the initial training 2D or 3D representation;
• and
• the plurality of sets of image data is from a plurality of different subjects.

A radiation therapy system comprising:

• a radiation therapy machine comprising a radiation source configured to direct a radiation beam into a subject positioned on a couch;
• a surface scanning system comprising a light detector configured to take a surface scan of the subject positioned on the couch;
• at least one memory configured to store:
• a computer-implemented tumor position determining model trained based on a plurality of sets of image data, each set of image data comprises an initial training image of a tumor in a subject, an initial training two-dimensional (2D) or three-dimensional (3D) representation of a surface of the subject, a subsequent training image of the tumor in the subject and a subsequent training 2D or 3D representation of a surface of the subject, wherein
• the subsequent training image and the subsequent training 2D or 3D representation are taken at a subsequent point in time than the initial training image and the initial training 2D or 3D representation;
• and
• the plurality of sets of image data is from a plurality of different subjects;
• and
• a medical image of the subject positioned on a couch in connection with a radiation therapy simulator;
• and
• at least one processor configured to:
• process the medical image and generate an initial image of a tumor in the subject and an initial 2D or 3D representation of a surface of the subject based on the medical image;
• process the surface scan and generate a subsequent 2D or 3D representation of a surface of the subject based on the surface scan;
• and
• determine a subsequent position of the tumor in the subject based on the subsequent 2D or 3D representation, the initial 2D or 3D representation, the initial image and the computer-implemented tumor position determining model.

A radiation therapy system comprising:

• a radiation therapy machine comprising a radiation source configured to direct a radiation beam into a subject positioned on a couch;
• a first surface scanning system comprising a light detector configured to take an initial surface scan of the subject positioned on a couch in connection with a medical imaging machine;
• a second surface scanning system comprising light detector configured to take a subsequent surface scan of the subject positioned on the couch in connection with the radiation therapy machine;
• at least one memory configured to store:
• a computer-implemented tumor position determining model trained based on a plurality of sets of image data, each set of image data comprises an initial training image of a tumor in a subject, an initial training two-dimensional (2D) or three-dimensional (3D) representation of a surface of the subject, a subsequent training image of the tumor in the subject and a subsequent training 2D or 3D representation of a surface of the subject, wherein
• the subsequent training image and the subsequent training 2D or 3D representation are taken at a subsequent point in time than the initial training image and the initial training 2D or 3D representation;
• and
• the plurality of sets of image data is from a plurality of different subjects;
• and
• an initial image of a tumor in the subject positioned on the couch in connection with the medical imaging machine;
• and
• at least one processor configured to:
• process the initial surface scan and generate an initial 2D or 3D representation of a surface of the subject based on the initial surface scan;
• process the subsequent surface scan and generate a subsequent 2D or 3D representation of a surface of the subject based on the subsequent surface scan;
• and
• determine a subsequent position of the tumor in the subject based on the subsequent 2D or 3D representation, the initial 3D representation, the initial image and the computer-implemented tumor position determining model.