Systems and methods for obtaining 3-d images from x-ray information

A method for obtaining a 3-dimensional image using a conventional 2-dimensional X-ray image, the method comprising:

• acquiring a first X-ray image of a bone in a first view;
• performing a 3-dimensional reconstruction with respect to the first view using the to obtain a first 3-dimensional reconstructed image;
• acquiring a second X-ray image of the bone in a second view;
• and
• performing a 3-dimensional reconstruction with respect to the second view using the first 3-dimensional reconstructed image as a template and to obtain a second 3-dimensional reconstructed image, wherein the performings each include, for the corresponding X-ray image,
• determining a camera model, of the X-ray image, wherein the determining uses known parameters to determine spatial values of a source and the bone,
• extracting a contour of the bone from the image, wherein the contour includes distinct anatomical regions,
• identifying anatomical values of the contour, wherein the anatomical values are 2-dimensional anatomical values from the distinct anatomical regions,
• importing a 3-dimensional template, template anatomical values, and template anatomical values, all corresponding to the bone,
• extracting silhouette vertices and silhouette vertex projections of the 3-dimensional template based on the camera model and an initial alignment of the 3-dimensional template,
• aligning the 3-dimensional template with respect to the input X-ray image,
• projecting the 3-dimensional template on to an acquired image plane, using the camera model, to obtain a 2-dimensional projection model,
• modifying the aligned template to match the 2-dimensional anatomical values,
• determining a best matching point on the extracted contour, for each extracted silhouette vertex projection, for 2-dimensional to 2-dimensional correspondence of each silhouette vertex projection to the extracted contour,
• back-projecting each of the best matching points according to the camera model to form a back projected ray, the ray being formed by the X-ray source and the best matching point,
• determining target positions, wherein the target positions are a closest position to a corresponding silhouette vertex on each of the back projected rays, and
• deforming the 3-dimensional template such that the extracted silhouette vertices achieve the target positions to obtain a 3-dimensional reconstructed image.

A system for obtaining a 3-dimensional image using at least one conventional 2-dimensional X-ray image, the system comprising:

• an inputter configured to acquire an X-ray image of a bone;
• a camera model determinator configured to determine a camera model, of the X-ray image, wherein the determining uses known parameters to determine spatial values of a source and the bone;
• a contourer configured to extract a contour of the bone from the image, wherein the contour includes distinct anatomical regions and identify anatomical values of the contour, wherein the anatomical values are 2-dimensional anatomical values from the distinct anatomical regions;
• an importer configured to import a 3-dimensional template, template anatomical values, and template anatomical values, all corresponding to the bone;
• a silhouette vortex extractor configured to extract silhouette vertices and silhouette vertex projections of the 3-dimensional template based on the camera model and an initial alignment of the 3-dimensional template;
• an aligner configured to align the 3-dimensional template with respect to the input X-ray image;
• a first estimator configured to project the 3-dimensional template on to an acquired image plane, using the camera model, to obtain a 2-dimensional projection model;
• a selective anatomical modifier configured to modify the aligned template to match the 2-dimensional anatomical values;
• a second estimator configured to determine a best matching point on the extracted contour, for each extracted silhouette vertex projection, for 2-dimensional to 2-dimensional correspondence of each silhouette vertex projection to the extracted contour, back-project each of the best matching points according to the camera model to form a back projected ray, the ray being formed by the X-ray source and the best matching point, and determine target positions, wherein the target positions are a closest position to a corresponding silhouette vertex on each of the back projected rays;
• and
• a deformer configured to deform the 3-dimensional template such that the extracted silhouette vertices achieve the target positions to obtain a 3-dimensional reconstructed image.