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 texture features for response monitoring of glioblastoma multiforme on -weighted and -FLAIR MR images: A preliminary investigation in terms of identification and segmentation
Rent:
Rent this article for
USD
10.1118/1.3357289
/content/aapm/journal/medphys/37/4/10.1118/1.3357289
http://aip.metastore.ingenta.com/content/aapm/journal/medphys/37/4/10.1118/1.3357289

Figures

Image of FIG. 1.
FIG. 1.

A rectangular ROI encompassing a tumor region (delineated outline) in the frontal lobe taken from a -weighted contrast-enhanced (FSPGR) MR image slice of a GBM patient (left) and its power map computed using Eq. (16) (right).

Image of FIG. 2.
FIG. 2.

-weighted contrast-enhanced (FSPGR) MRI of the brain (top row) and corresponding -FLAIR MRI (bottom row) for three different patients after partial resection and prior to radiotherapy. The last column shows a case with multifocal disease. A radiation oncologist delineated the gross tumor volume on the -weighted images corresponding to the contrast-enhancing lesion after IV administration of gadolinium DTPA contrast. After performing rigid registration based on a normalized mutual information, all contours on the -weighted images were applied to the -FLAIR images.

Image of FIG. 3.
FIG. 3.

A selected ROI and its texture map: (left) A rectangular ROI with a tumor at the center taken from a -weighted contrast-enhanced (FSPGR) MRI, and (right) a contour plot of the texture map of the Variance feature computed from the GLCM of the power map of the -weighted image. For the texture map, black corresponds to larger feature values. The white dotted rectangle enclosing the tumor represents the region used for the texture based identification as described in Sec. IV B.

Image of FIG. 4.
FIG. 4.

Sum-mean feature computed for brain tissue (filled circles) and tumor (open circles) ROIs on -weighted contrast-enhanced (FSPGR) MR images for all 27 patients, computed from the GLCM of the power map (left), computed from the GLCM of the original ROI on the -weighted image (middle), and mean descriptor (right). The 27 feature values for the brain tissue and the 27 values for the tumors are both sorted in ascending order.

Image of FIG. 5.
FIG. 5.

Variance feature computed for brain tissue (filled circles) and tumor (open circles) ROIs on -weighted contrast-enhanced (FSPGR) MR images for all 27 patients; computed from the GLCM of the power map (left), computed from the GLCM of the original ROI on the -weighted image (middle), and mean descriptor (right, same as Fig. 4). The 27 feature values for the brain tissue and the 27 values for the tumors are both sorted in ascending order.

Image of FIG. 6.
FIG. 6.

Sum-mean feature computed for brain tissue (filled circles) and tumor (open circles) ROIs on -FLAIR MR images for all 27 patients, computed from the GLCM of the power map (left), computed from the GLCM of the original ROI on the -FLAIR image (middle), and mean descriptor (right). The 27 feature values for the brain tissue and the 27 values for the tumors are both sorted in ascending order.

Image of FIG. 7.
FIG. 7.

Variance feature computed for brain tissue (filled circles) and tumor (open circles) ROIs on -FLAIR MR images for all 27 patients, computed from the GLCM of the power map (left), computed from the GLCM of the original ROI on the -FLAIR image (middle), and mean descriptor (right, same as Fig. 6). The 27 feature values for the brain tissue and the 27 values for the tumors are both sorted in ascending order.

Image of FIG. 8.
FIG. 8.

ROC curves plotted for median Sum-mean texture features computed for brain tissue and tumor -weighted contrast-enhanced (FSPGR) MR ROIs taken from all 27 patients; feature computed from the GLCM of the power map (left), GLCM of the original ROI (middle), and mean descriptor (right).

Image of FIG. 9.
FIG. 9.

Normalized texture maps using five intensity levels (black: Highest level, white: Smallest level) for three methods: Variance feature map computed from the GLCM of the power map of the ROI in the -weighted image (left), Variance feature map computed from the GLCM of the ROI in the original -weighted image (middle), and the mean value descriptor (right). The bold dark line is the contour drawn by the radiation oncologist. The OI values for the images in the first row were 0.861, 0.875, and 0.872 and for the images in the second row were 0.853, 0.852, and 0.855 for the , , and the mean value descriptor, respectively.

Image of FIG. 10.
FIG. 10.

Segmentation power values computed to differentiate a tumor from the surrounding tissue based on the Variance second order feature; computed from the GLCM of the power map (light gray solid), from the GLCM of original ROI (dark gray solid), and from the mean value (dashed-dotted).

Tables

Generic image for table
TABLE I.

AUC values, maximum classification accuracies, and Canberra distances for features computed on brain tissues and tumors using the GLCM of the power map , the GLCM of the original ROI , and the mean value descriptor.

Loading

Article metrics loading...

/content/aapm/journal/medphys/37/4/10.1118/1.3357289
2010-03-25
2014-04-24
Loading

Full text loading...

This is a required field
Please enter a valid email address
752b84549af89a08dbdd7fdb8b9568b5 journal.articlezxybnytfddd
Scitation: Robust texture features for response monitoring of glioblastoma multiforme on T1-weighted and T2-FLAIR MR images: A preliminary investigation in terms of identification and segmentation
http://aip.metastore.ingenta.com/content/aapm/journal/medphys/37/4/10.1118/1.3357289
10.1118/1.3357289
SEARCH_EXPAND_ITEM