X-ray system for the iterative determination of an optimal coordinate transformation between overlapping volumes that have been reconstructed from volume data sets of discretely scanned object areas

Inventors

König, Thomas • Hörndler, Klaus

Assignees

Ziehm Imaging GmbH

Abstract

This disclosure relates to an X-ray system for obtaining referencing information in the form of a coordinate transformation between a first volume and a second volume, with the volumes partially overlapping. In this case two partially overlapping subareas of an extended object having a continuous distribution of a physical size are discretely sampled with a measuring apparatus; the volumes are reconstructed and the overlapping volumes are iteratively changed, wherein a bandpass filter is applied to the overlapping volumes, where said bandpass filter changes the Fourier domain representation of the overlapping volumes. The filtered overlapping volumes, in which the artifacts, induced by the discrete sampling, are matched, are compared by means of a measure of similarity, which is iteratively maximized by means of mathematical optimization. The coordinate transformation with the maximum result for the measure of similarity is used as the optimal coordinate transformation for combining the two volumes in the correct position. One application is the combination of 3D X-ray volumes that have been reconstructed by means of a cone beam computed tomography scanner, preferably a 3D C-arm.

Core Innovation

The invention is an X-ray system suitable for iterative determination of an optimal coordinate transformation between a first N-dimensional partial volume V1 and a second N-dimensional partial volume V2 of an extended object, where the partial volumes partially overlap in overlapping volumes OV1 and UOV2. The system records the partial volumes by discrete sampling using an imaging device. During iterative steps i, the overlapping volumes are filtered with a bandpass filter to obtain filtered overlapping volumes FOV1(i) and FOV2(i).

In the bandpass filtering, frequencies that are not contained in the overlapping region are removed from each overlapping volume, including removing from OV1(i) frequencies that are not contained in UOV2(i), and removing from UOV2(i) frequencies that are not contained in OV1(i). This filtering provides filtered overlapping volumes for subsequent comparison and supports determining the optimal coordinate transformation by iterative optimization. The optimization uses a measure of similarity between the filtered overlapping volumes FOV1(i) and FUOV2(i) across the plurality of iteration steps.

Once the optimal coordinate transformation is determined, the invention registers and stitches the partial volumes into a composite volume using the best transformation. Dependent features describe that the partial volumes may be aligned using rigid and elastic registration and merged using a weighted voxel averaging/transition function to reduce cone-beam transition artifacts. The system can store referencing information for later use, including referencing between the C-arm and the object support structure, to enable later coordinate use.

Claims Coverage

The independent claim specifies an X-ray system that iteratively records two partially overlapping N-dimensional partial volumes, bandpass-filters their overlapping regions to remove non-common frequencies, and determines the optimal coordinate transformation by iteratively maximizing a measure of similarity between the filtered overlapping volumes. The provided partial content and claim list contain only one independent claim (clm-00001).

For the independent claim present, the core claim coverage lies in the iterative loop combining bandpass filtering of partially overlapping N-dimensional volumes by removing non-overlapping frequencies and iterative optimization of a similarity measure computed on the filtered overlapping volumes to obtain the optimal coordinate transformation.

Stated Advantages

Documented Applications