Alignment of mixed-modality data sets for reduction and removal of imaging artifacts
Methods and systems are described for removing reflective artifacts from an imaging. A first volumetric model and a second volumetric model of the same object are accessed from a computer-readable memory. The orientation and scale of at least one of the two models is repeatedly and automatically adjusted until an optimized orientation and scale is determined that correlates the first volumetric model and the second volumetric model. The second volumetric model is then overlaid onto the first volumetric model. Any data points in the first volumetric model that extend beyond a surface of the second volumetric model are detected and removed to create an artifact-reduced volumetric model.
1. A method for removing artifacts from an imaging model, the method comprising:
accessing a first volumetric model and a second volumetric model;
repeatedly automatically adjusting an orientation of the first volumetric model or the second volumetric model until an optimized orientation is determined that correlates the first volumetric model and the second volumetric model,
wherein repeatedly automatically adjusting the orientation of the first volumetric model or the second volumetric model includes applying a simulated annealing routine to adjust the orientation of the first volumetric model or the second volumetric model;
overlaying the second volumetric model onto the first volumetric model;
detecting data points in the first volumetric model that extend beyond a surface of the second volumetric model; and
adjusting the detected data point from the first volumetric model to create an artifact-reduced volumetric model.
2. The method of claim 1 , wherein applying the simulated annealing routine includes
evaluating an alignment of the first volumetric model and the second model,
adjusting a variable of the first volumetric model or the second volumetric model, the variable being randomly selected from a group consisting of a yaw, a pitch, and a roll,
evaluating an alignment of the first volumetric model and the second volumetric model after adjusting the variable,
accepting the adjustment of the variable if the alignment of the first volumetric model and the second volumetric model is improved after the adjustment of the variable,
accepting the adjustment of the variable if the alignment of the first volumetric model and the second volumetric model is not improved after the adjustment of the variable and a calculated acceptance probability exceeds a threshold, and
rejecting the adjustment of the variable if the alignment of the first volumetric model and the second volumetric model is not improved after the adjustment of the variable and the calculated acceptance probability does not exceed the threshold.
3. The method of claim 2 , wherein repeatedly automatically adjusting the orientation of the first volumetric model or the second volumetric model includes repeating the acts of the simulated annealing routine.
4. The method of claim 3 , wherein applying the simulated annealing routine includes reducing a value of a T variable each time the simulated annealing routine is repeated, and wherein reducing the value of the T variable reduces the likelihood that the calculated acceptance probability will exceed the threshold.
5. The method of claim 3 , wherein applying the simulated annealing routine includes generating a random number each time the simulated annealing routine is repeated, and wherein determining whether the calculated acceptance probability exceeds the threshold includes determining whether the calculated acceptance probability exceeds the generated random number.
6. The method of claim 1 , wherein the second volumetric model includes an optical surface scan model of a structure.
7. The method of claim 6 , wherein repeatedly automatically adjusting an orientation of the first volumetric model or the second volumetric model includes repeatedly automatically adjusting the orientation of the second volumetric model and maintaining a constant orientation of the first volumetric model.
8. The method of claim 7 , wherein the first volumetric model includes a CT model of the structure.
9. The method of claim 6 , further comprising capturing surface scan data using a laser scanning system; and generating the surface scan model of the structure based on the captured surface scan data.
10. The method of claim 1 , wherein detecting data points in the first volumetric model that extend beyond the surface of the second volumetric model includes determining that all data points from the first volumetric model that extend beyond the surface of the second volumetric model represent artifacts in the first volumetric model.
11. The method of claim 1 , wherein adjusting the detected data point from the first volumetric model to create an artifact-reduced volumetric model includes adjusting the detected data points by interpolation of the detected data points.
12. A system for removing artifacts from an imaging model, the system comprising:
an x-ray source;
an x-ray detector that captures x-ray images;
a surface scanner that captures a surface scan; and
an image processor that constructs a first volumetric model and a second volumetric model, wherein the first volumetric model and the second volumetric model include a three-dimensional CT model of a structure from the x-ray images and a three-dimensional surface model of the structure from the surface scan, the image processor configured to
repeatedly automatically adjust an orientation of the first volumetric model or the second volumetric model until an optimized orientation is determined that correlates the first volumetric model and the second volumetric model;
overlay the second volumetric model onto the first volumetric model;
detect data points in the first volumetric model that extend beyond a surface of the structure in the second volumetric model; and
adjust the detected data point from the first volumetric model to create an artifact-reduced volumetric model,
wherein the image processor is configured to repeatedly automatically adjust the orientation of the three-dimensional CT model or the surface model by applying a simulated annealing routine to adjust the orientation of the first volumetric model or the second volumetric model.
13. The system of claim 12 , wherein the image processor is configured to apply the simulated annealing routine by
evaluating an alignment of the first volumetric model and the second model,
adjusting a variable of the first volumetric model or the second volumetric model, the variable being randomly selected from a group consisting of a yaw, a pitch, and a roll,
evaluating an alignment of the first volumetric model and the second volumetric model after adjusting the variable,
accepting the adjustment of the variable if the alignment of the first volumetric model and the second volumetric model is improved after the adjustment of the variable,
accepting the adjustment of the variable if the alignment of the first volumetric model and the second volumetric model is not improved after the adjustment of the variable and a calculated acceptance probability exceeds a threshold, and
rejecting the adjustment of the variable if the alignment of the first volumetric model and the second volumetric model is not improved after the adjustment of the variable and the calculated acceptance probability does not exceed the threshold.
14. The system of claim 13 , wherein the image processing system is configured to repeatedly automatically adjust the orientation of the three-dimensional CT model or the surface model by repeating the acts of simulated annealing routine.
15. The system of claim 12 , wherein the image processor is configured to detect data points in the first volumetric model that extend beyond a surface of the structure in the second volumetric model by determining that all data points from the first volumetric model that extend beyond the surface of the structure in the second volumetric model represent artifacts in the first volumetric model.
16. The system of claim 12 , wherein the image processor is configured to adjust the detected data point from the first volumetric model to create an artifact-reduced volumetric model by adjusting the detected data points by interpolation of the detected data points.
17. A method of automatically aligning a first volumetric model and a second volumetric model, the method comprising repeating the acts of:
evaluating an alignment of the first volumetric model and the second volumetric model,
adjusting a variable of the first volumetric model or the second volumetric model, the variable being randomly selected from a group consisting of a yaw, a pitch, a roll, and a scale,
evaluating an alignment of the first volumetric model and the second volumetric model after adjusting the variable,
accepting the adjustment of the variable if the alignment of the first volumetric model and the second volumetric model is improved after the adjustment of the variable,
generating a random threshold number;
accepting the adjustment of the variable if the alignment of the first volumetric model and the second volumetric model is not improved after the adjustment of the variable and a calculated acceptance probability exceeds the random threshold number,
rejecting the adjustment of the variable if the alignment of the first volumetric model and the second volumetric model is not improved after the adjustment of the variable and the calculated acceptance probability does not exceed the threshold number; and
adjusting a probability variable used to calculate the acceptance probability, wherein adjusting the probability variable reduces the likelihood that the calculated acceptance probability will exceed the random threshold number on each subsequent repeat iteration.