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.
Improving image-guided radiation therapy of lung cancer by reconstructing 4D-CT from a single free-breathing 3D-CT on the treatment day
Rent:
Rent this article for
USD
10.1118/1.4768226
/content/aapm/journal/medphys/39/12/10.1118/1.4768226
http://aip.metastore.ingenta.com/content/aapm/journal/medphys/39/12/10.1118/1.4768226

Figures

Image of FIG. 1.
FIG. 1.

The overview of our proposed method for estimating the 4D-CT images in the treatment day (d) from a free-breathing image in the treatment day (c), by using the 4D-CT model built in the planning day (a). Our method first deinterlaces the free-breathing image into the respective phase images (b), and then reconstructs a new 4D-CT in the treatment day (d) by warping the planning day 4D-CT onto the deinterlaced images, with the guidance of the 4D-CT model built in the planning day.

Image of FIG. 2.
FIG. 2.

The schematic illustration of our spatiotemporal registration on 4D-CT. The deformation fields {f s } are estimated based on the correspondences between key points by robust feature matching with respect to their shape and appearance. Meanwhile, the temporal continuity is preserved by performing kernel regression along the temporal fiber bundles Φ (see the blue dashed curves).

Image of FIG. 3.
FIG. 3.

The pipeline of reconstructing 4D-CT in the treatment day. First, the free-breathing 3D-CT is roughly aligned with planning day 4D-CT by bone alignment. Then, the mixed phase information in I will be deinterlaced into partial 4D-CT D according to the respiratory motion. Next, 4D-to-4D image registration is performed between the partial 4D-CT D and the complete 4D-CT P to reconstruct the treatment day 4D-CT T by iteratively establishing spatial correspondences on key points, propagating correspondences along the respiratory motion, interpolating the dense deformation field, and adjusting the treatment day respiratory motion.

Image of FIG. 4.
FIG. 4.

The result of bone alignment. (a) shows the fusion of bone in treatment-day free-breathing 3D-CT (with smaller rib cage) over the particular phase image (with longer rib cage) before bone alignment. (b) shows the same fusion result after bone alignment. It is clear that the bones in and are approximately registered.

Image of FIG. 5.
FIG. 5.

The scheme of constructing partial 4D-CT D from a free-breathing 3D-CT I by phase deinterlace. First, the entire image I is portioned into a number of local 2D patches by Oct-tree technique. Then we disseminate each local patch θ I (x, l) to the particular phase image , which hold the best matched local patch with respect to θ I (x, l).

Image of FIG. 6.
FIG. 6.

The scheme of adjusting the treatment day respiratory motion between any phases s and t.

Image of FIG. 7.
FIG. 7.

The performance of 4D-CT reconstruction from a simulated free-breathing 3D-CT. A planning day 4D-CT is first obtained, with its maximum inhale, middle, and maximum exhale phases shown in (a). The simulated treatment day 4D-CT is displayed in (b), along with a simulated free-breathing 3D-CT of the treatment day shown in (c). The registration results between the planning day 4D-CT and the free-breathing 3D-CT by diffeomorphic Demons and SyN are shown in (d) and (f), with their difference images with respect to ground-truth (b) displayed in (e) and (g), respectively. The reconstruction results by our method without/with temporal guidance are displayed in (h) and (i), along with their corresponding difference images with respect to the ground-truth (b) shown in (i) and (k), respectively.

Image of FIG. 8.
FIG. 8.

The CT images scanned in a typical subject during lung cancer radiation therapy. The 4D-CT (with 6 phase images P 1, …, P 6) scanned in the planning-day and the free-breathing 3D-CT images taken in the five treatment days are displayed in the top and middle panels of figure, respectively, with tumor also delineated by contours. Three consecutive slices of the second-treatment-day free-breathing 3D-CT are shown in the bottom for demonstrating the inconsistency for the tumor regions (indicated by arrows).

Image of FIG. 9.
FIG. 9.

Evaluation of the reconstructed 4D-CT from a single free-breathing 3D-CT in the planning day. The 4D-CT P and free-breathing 3D-CT in the planning day are shown in (a) and (b), respectively. The reconstructed 4D-CT T by diffeomorphic Demons and our method are shown in (c) and (e), with their difference images with respect to the planning day 4D-CT phase images displayed in (d) and (f), respectively.

Image of FIG. 10.
FIG. 10.

Reconstructed 4D-CT from a single free-breathing 3D-CT in the treatment day. The free-breathing 3D-CT in the treatment day is shown in (a). The reconstructed 4D-CTs from (a) by bone alignment, diffeomorphic Demons, and our reconstruction method are displayed in (b), (c), and (d), respectively.

Image of FIG. 11.
FIG. 11.

The overlap of tumor contour drawn on the free-breathing 3D-CT and the tumor movement field in the reconstructed treatment-day 4D-CT. The shape of tumor in the treatment-day free-breathing 3D-CT and the tumor movement field in the reconstructed treatment-day 4D-CT is displayed by small surface and large surface, respectively.

Image of FIG. 12.
FIG. 12.

Reconstructed 4D-CT from the single free-breathing 3D-CT in three treatment days. The tumors in the treatment-day free-breathing 3D-CT image are manually delineated by the inner contour. The estimated tumor movement fields in the treatment days are outlined by the outer contour.

Tables

Generic image for table
TABLE I.

Summary of important notations.

Generic image for table
TABLE II.

The mean and standard deviation of registration errors (mm) on 300 landmark points between maximum inhale and exhale phases.

Generic image for table
TABLE III.

The mean and standard deviation of registration errors (mm) on 75 landmark points across all six phases.

Loading

Article metrics loading...

/content/aapm/journal/medphys/39/12/10.1118/1.4768226
2012-12-03
2014-04-23
Loading

Full text loading...

This is a required field
Please enter a valid email address
752b84549af89a08dbdd7fdb8b9568b5 journal.articlezxybnytfddd
Scitation: Improving image-guided radiation therapy of lung cancer by reconstructing 4D-CT from a single free-breathing 3D-CT on the treatment day
http://aip.metastore.ingenta.com/content/aapm/journal/medphys/39/12/10.1118/1.4768226
10.1118/1.4768226
SEARCH_EXPAND_ITEM