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.
Robust surface registration using salient anatomical features for image-guided liver surgery: Algorithm and validation
Rent:
Rent this article for
USD
10.1118/1.2911920
/content/aapm/journal/medphys/35/6/10.1118/1.2911920
http://aip.metastore.ingenta.com/content/aapm/journal/medphys/35/6/10.1118/1.2911920

Figures

Image of FIG. 1.
FIG. 1.

Example of poor initial alignment (a),(b) and resulting misregistration (c),(d) of clinical data obtained using a traditional ICP algorithm. Note that the LRS scan of the anterior surface of the liver is registered to the posterior liver surface via ICP.

Image of FIG. 2.
FIG. 2.

Anatomical schematic (left) and examples of preoperative image (middle) and intraoperative LRS liver data (right) with corresponding falciform ligament regions outlined. Note that the falciform ligament region can be located on the preoperative image surface via the groove in the surface and texture can be used to delineate the falciform region in the LRS surface.

Image of FIG. 3.
FIG. 3.

Graphical depiction of the weighted point correspondence method. Only the Euclidean distances computed from source patch point to target patch point are biased by the weighting factor (i.e., dashed lines). The point correspondence determination for nonpatch points is not affected by the weighting (i.e., solid lines). Note that the graphical depiction represents the case where only a single patch region is used.

Image of FIG. 4.
FIG. 4.

Digital photograph (left), raw LRS scan (center) and sample CT slice (right) of imaging phantom. The silicon liver model, located in the center of the phantom, is surrounded by a set of seven white Teflon spheres. These spheres, which can be localized in both LRS and CT image spaces, are used in the determination of the gold standard ICP registration and serve as targets in the robustness studies.

Image of FIG. 5.
FIG. 5.

Phantom LRS simulated falciform patch selected from full scan (left) was used to delineate the homologous region in the CT image surface (center). To more accurately simulate the typical LRS surface field of view obtained during surgery, a subregion of the LRS was manually selected for use in the robustness trials (right).

Image of FIG. 6.
FIG. 6.

Traditional ICP registration results (a) and overlaid image and falciform patch regions (b) for the clinical data used in the robustness trials (Patient 3). The falciform and inferior ridge regions delineated in the intraoperative LRS and preoperative CT data are shown in panels (c) and (d), respectively. Note the large contrast in the accuracy of the alignment in this case than that shown in Fig. 1. The RMS residual for this registration was .

Image of FIG. 7.
FIG. 7.

Clinical results for Patient 1 showing visualizations of the ICP registration (a–b) and patch ICP registration using a single (falciform) patch (c–d) initialized using the anatomical fiducial PBR, as well as patch ICP registration using multiple patches (falciform and inferior ridge) given no initial alignment registration (e–f). The LRS and ICP patches are highlighted for the ICP and patch ICP registrations in (b,d,f). The ICP registration shows an apparent misalignment which is corrected via the proposed method. For reference, the number of points containing the preoperative liver, falciform, and inferior ridge regions derived from CT images were , 3074, and 1522, respectively. The number of points in the LRS scan representation of the liver, falciform, and inferior ridge regions for this clinical dataset were , , and 605, respectively.

Image of FIG. 8.
FIG. 8.

Clinical results for Patient 5 showing visualizations of the ICP registration (a–b) and the patch ICP registration using a single (falciform) patch (c–d) initialized using the anatomical fiducial PBR, as well as patch ICP registration using multiple patches (falciform and inferior ridge) given no initial alignment registration (e–f). The LRS and ICP patches are highlighted for the ICP and patch ICP registrations in (b, d, f). For reference, the number of points containing the preoperative liver, falciform, and inferior ridge regions derived from CT images were , 3139, and l548, respectively. The number of points in the LRS scan representation of the liver, falciform, and inferior ridge regions for this clinical dataset were , 923, and 358, respectively.

Tables

Generic image for table
TABLE I.

Summary of results for the small scale perturbation robustness trials using the phantom dataset shown in Fig. 5. The number of successful trials (out of 250), RMS residual and TRE overall trails, and RMS residual and TRE over successful trials are reported for each registration method. A successful trial is determined as that which yields a TRE of less than . For reference, the gold standard ICP registration for the phantom yielded RMS residual and TRE values of and , respectively.

Generic image for table
TABLE II.

Summary of results for the large scale perturbation robustness trials using the phantom dataset shown in Fig. 5. The number of successful trials (out of 250), RMS residual and TRE over all trails, and RMS residual and TRE over successful trials is reported for each registration method. A successful trial is determined as that which yields a TRE of less than . For reference, the gold standard ICP registration for the phantom yielded RMS residual and TRE values of 0.6 and , respectively.

Generic image for table
TABLE III.

Summary of the registration results for the six clinical datasets using no initial alignment transformation. The results are shown for the ICP, patch ICP registration with a single feature (PICP), and patch ICP registration with multiple features (PICP2) in terms of the RMS residual between the entire surfaces as well as the homologous patch regions. Feature 1 represents the falciform ligament region and feature 2 denotes the inferior ridge region. Grossly misaligned registrations are noted with a superscript and were determined by visual inspection.

Generic image for table
TABLE IV.

Summary of the registration results for the six clinical datasets using the anatomical fiducial-based PBR initial alignment. The results are shown for the ICP, patch ICP registration with a single feature (PICP), and patch ICP registration with multiple features (PICP2) in terms of the RMS residual between the entire surfaces as well as the homologous patch regions. Feature 1 represents the falciform ligament region and feature 2 denotes the inferior ridge region. Grossly misaligned registrations are noted with a superscript and were determined by visual inspection.

Generic image for table
TABLE V.

Summary of results for the small scale perturbation robustness trials using the clinical dataset shown in Fig. 6. The number of successful trials (out of 250), mean residual over all trails, and mean residual over successful is reported for each registration method. A successful trial is determined as that which yields a RMS residual of less than over the entire surface. For reference, the gold standard ICP registration (shown in Fig. 6) yielded an RMS residual of .

Generic image for table
TABLE VI.

Summary of results for the large scale perturbation robustness trials using the clinical dataset shown in Fig. 6. The number of successful trials (out of 250), mean residual over all trails, and mean residual over successful trials is reported for each registration method. A successful trial is determined as that which yields a RMS residual of less than over the entire surface. For reference, the gold standard ICP registration (shown in Fig. 6) yielded an RMS residual of .

Generic image for table
TABLE VII.

Comparative summary of the time to solution of each algorithm under the condition of large scale and small scale perturbations for both phantom and clinical datasets. The reported solution times were averaged over the successful registration runs for each trial and reported both as mean total time as well as mean time per iteration for each algorithm.

Loading

Article metrics loading...

/content/aapm/journal/medphys/35/6/10.1118/1.2911920
2008-05-28
2014-04-24
Loading

Full text loading...

This is a required field
Please enter a valid email address
752b84549af89a08dbdd7fdb8b9568b5 journal.articlezxybnytfddd
Scitation: Robust surface registration using salient anatomical features for image-guided liver surgery: Algorithm and validation
http://aip.metastore.ingenta.com/content/aapm/journal/medphys/35/6/10.1118/1.2911920
10.1118/1.2911920
SEARCH_EXPAND_ITEM