Method and system for in situ targeting of objects
A method for tracking a target located within a body, whilst scanning the body using X-ray projections, the method including the steps of: (a) providing a first prior image of the body and target indicating their relative position; (b) iteratively performing, for a series of time intervals, the steps of: (i) acquiring a few spatially distributed x-ray projections of the target area; (ii) determining one of a series of motion phases for each x-ray projection; (iii) performing an iterative reconstruction of the image for each of the motion phases; and (iv) outputting a resultant image for a current time interval.
1. A method of predicting a 3D target position of a moving in-situ target in a body, the method comprising:
(a) providing a computational prior model of a moving in-situ target in a body and the surrounding environment of the moving in-situ target;
(b) selecting a respiratory phase of the computational prior model using a surrogate signal of respiratory motion to provide a computational prior model of the moving in-situ target and the surrounding environment of the moving in-situ target at a selected respiratory phase;
(c) dividing the computational prior model of the moving in-situ target and the surrounding environment of the moving in-situ target at the selected respiratory phase to provide: 1) a target model of the moving in-situ target; and 2) an anatomy-only model of anatomies surrounding the moving in-situ target, wherein said anatomy-only model does not provide a model of said moving in-situ target;
(d) selecting one or more spatially distributed real-time 2D image of the moving in-situ target at the selected respiratory phase using the surrogate signal of respiratory motion;
(e) comparing the anatomy-only model to the one or more spatially distributed real-time 2D image; and
(f) minimizing a difference between the anatomy-only model and the one or more spatially distributed real-time 2D image to predict the 3D target position of the moving in-situ target in a body.
2. The method as claimed in claim 1 , wherein the moving in-situ target comprises a tumor and dividing the computational prior model provides a model of the tumor and an anatomy-only model of anatomies surrounding the tumor based on a pre-treatment tumor contour formed around the tumor.
3. The method as claimed in claim 1 , wherein said method is applied to tumor tracking, and wherein comparing the anatomy-only model to the one or more spatially distributed real-time 2D image comprises subtracting the anatomy-only model from the one or more spatially distributed real-time 2D image.
4. The method as claimed in claim 1 , wherein said method is applied to tumor tracking, and comparing the anatomy-only model to the one or more spatially distributed real-time 2D image comprises minimizing additional regularization terms that penalize large spatial deviations of the moving in-situ target from a prior position.
5. The method as claimed in claim 4 , wherein the prior position is the moving in-situ target position measured in the computational prior model.
6. The method as claimed in claim 1 , wherein said computational prior model comprises a pre-treatment 4D cone-beam CT image.
7. A method of predicting a 3D position of a moving in-situ tumor in a body, the method comprising:
(a) providing a computational prior model of a moving in-situ tumor in a body and anatomies surrounding the moving in-situ tumor;
(b) selecting a respiratory phase of the computational prior model using a surrogate signal of respiratory motion to provide a computational prior model at the selected respiratory phase of the moving in-situ tumor and anatomies surrounding the moving in-situ tumor;
(c) removing said moving in-situ tumor from said computational prior model at the selected respiratory phase to provide an anatomy-only model;
(d) selecting one or more spatially distributed real-time 2D image of the moving in-situ tumor at the selected respiratory phase;
(e) comparing the anatomy-only model to the one or more spatially distributed real-time 2D image; and
(f) minimizing a difference between the anatomy-only model and the one or more spatially distributed real-time 2D image to predict the 3D position of the moving in-situ tumor in a body.