1887
banner image
No data available.
Please log in to see this content.
You have no subscription access to this content.
No metrics data to plot.
The attempt to load metrics for this article has failed.
The attempt to plot a graph for these metrics has failed.
Statistical shape model-based reconstruction of a scaled, patient-specific surface model of the pelvis from a single standard AP x-ray radiograph
Rent:
Rent this article for
USD
10.1118/1.3327453
/content/aapm/journal/medphys/37/4/10.1118/1.3327453
http://aip.metastore.ingenta.com/content/aapm/journal/medphys/37/4/10.1118/1.3327453

Figures

Image of FIG. 1.
FIG. 1.

A schematic diagram of different steps in the present single-image based 2D/3D reconstruction technique.

Image of FIG. 2.
FIG. 2.

The first two eigenmodes of the variation in our PDM of the pelvis. The shape instances were generated by evaluating with .

Image of FIG. 3.
FIG. 3.

Example of semiautomatic contour extraction and dominant point detection. (a) The contours extracted by the Livewire algorithm, where the crosses indicate the seed points; (b) a schematic diagram showing how the dominant points are detected; (c) dominant points (larger dots) extracted from the contours shown in (a); and (d) points used for the 2D/3D reconstruction, which are interpolated from the detected dominant points by a cubic spline.

Image of FIG. 4.
FIG. 4.

Schematic views of establishment of the projection geometry for a given AP x-ray radiograph and the anatomical landmarks used for algorithm initialization. (a) The radiograph coordinate system and the cone-beam projection model; (b) landmarks extracted from the mean model of the PDM; and (c) landmarks interactively extracted from the x-ray radiograph.

Image of FIG. 5.
FIG. 5.

Schematic views of the iterative landmark-to-ray registration. (a) A schematic view of data preparation step and (b) a schematic view of how to find 3D point pairs.

Image of FIG. 6.
FIG. 6.

Screenshots of establishing image-to-model correspondences. (a) The apparent contours of the mean model of the PDM after the landmark-based initialization and (b) 2D/2D correspondences between the image contours and the projections of the apparent contours of the mean model.

Image of FIG. 7.
FIG. 7.

Schematic illustration of computing 3D point pairs between a model and the input image from the established 2D/2D correspondences.

Image of FIG. 8.
FIG. 8.

Screenshots of different stages of our single-image based 2D/3D reconstruction technique. (a) The mean model of the PDM transformed by the iterative affine registration; (b) the reconstructed surface model; and (c) superposition of the projections of the apparent contours of the reconstructed surface models onto the x-ray radiograph.

Image of FIG. 9.
FIG. 9.

The five x-ray radiographs acquired when a randomly chosen cadaver pelvis was placed in different orientations with respect to the x-ray plate.

Image of FIG. 10.
FIG. 10.

Color-coded error distribution when the reconstructed surface model of the pelvis shown in Fig. 8 was compared to its associated ground truth after a surface-based iterative affine registration. (a) The front view and (b) the back view. In both images, the left column shows the color-coded errors; the middle column shows the ground truth surface model with a color-coded error distribution; and the right column shows the reconstructed surface model.

Image of FIG. 11.
FIG. 11.

Errors of reconstructing surface models of all 14 pelvises when a surface-based iterative affine registration was used to match the ground truth surface model to the associated reconstruction surface model. An average mean reconstruction error of 1.6 mm was observed.

Image of FIG. 12.
FIG. 12.

Errors of reconstructing surface models of all 14 pelvises when a surface-based iterative scaled rigid registration was used to match the ground truth surface model to the associated reconstruction surface model. An average mean reconstruction error of 1.9 mm was observed.

Image of FIG. 13.
FIG. 13.

Errors of reconstructing surface models of the 14 pelvises using the single-image based 2D/3D reconstruction technique that we introduced previously (Refs. 29 and 30). (a) Errors when a surface-based iterative affine registration was used to match the ground truth surface model to the associated reconstruction surface model and (b) errors when a surface-based iterative scaled rigid registration was used.

Image of FIG. 14.
FIG. 14.

Errors of reconstructing surface models of a randomly chosen cadaver pelvis when it was placed in five different orientations with respect to the x-ray plate. (a) Errors when a surface-based iterative affine registration was used to match the ground truth surface model to the associated reconstruction surface model and (b) errors when a surface-based iterative scaled rigid registration was used.

Tables

Generic image for table
TABLE I.

Estimated scales between the reconstructed surface model and the CT-derived ground truth along each coordinate axis.

Generic image for table
TABLE II.

Average scales between the reconstructed surface model and the CT-derived ground truth along each coordinate axis.

Generic image for table
TABLE III.

The paired two-tailed -test results when the estimated scales along different axes are compared to each other.

Loading

Article metrics loading...

/content/aapm/journal/medphys/37/4/10.1118/1.3327453
2010-03-09
2014-04-24
Loading

Full text loading...

This is a required field
Please enter a valid email address
752b84549af89a08dbdd7fdb8b9568b5 journal.articlezxybnytfddd
Scitation: Statistical shape model-based reconstruction of a scaled, patient-specific surface model of the pelvis from a single standard AP x-ray radiograph
http://aip.metastore.ingenta.com/content/aapm/journal/medphys/37/4/10.1118/1.3327453
10.1118/1.3327453
SEARCH_EXPAND_ITEM