RSNA 2005 

Abstract Archives of the RSNA, 2005


SSM17-04

Multi-scale Matched Mask Bone Elimination (multi-scale MMBE): Improved Automatic Bone Removal in Computed Tomography Angiography (CTA) Images

Scientific Papers

Presented on November 30, 2005
Presented as part of SSM17: Physics (CT: Cardiac Imaging II—Vascular)

Participants

Hugo A.F. Gratama van Andel MS, Presenter: Nothing to Disclose
Henk Willem Venema PhD, Abstract Co-Author: Nothing to Disclose
Geert J. Streekstra PhD, Abstract Co-Author: Nothing to Disclose
Marcel Van Straten PhD, Abstract Co-Author: Nothing to Disclose
Gerard J. Den Heeten MD, Abstract Co-Author: Nothing to Disclose
Cornelis A Grimbergen PhD, Abstract Co-Author: Nothing to Disclose

PURPOSE

To improve the quality of the mask that is used in an automatic method to remove bone from maximum intensity projection (MIP) and volume rendered (VR) images, using a multi-scale approach.

METHOD AND MATERIALS

For CTA-examinations with the MMBE method, a low dose scan is made before contrast injection. After matching the pre-contrast scan is converted into a mask, which is used to remove the bone in the CTA images (Radiology 2001;218:893-898). This improves visualization of the vessels in MIP and VR images. In this masking step vessels that lie close to the bone may be slightly affected at their edges. In the original method the scans are reconstructed with a smooth kernel, because of the requirements of the radiologists, and therefore the zone next to the bone that is also masked is relatively broad (≃ 1 mm). In the proposed method both scans are reconstructed with a sharp reconstruction kernel. The in-plane resolution is improved, a narrower bone mask is obtained and thin bone structures are masked as well. After masking the CTA-images are blurred with a Gaussian kernel in order to obtain CTA-images with resolution and noise level comparable to images reconstructed with the smooth kernel (IEEE TMI 2003;22:846-853). The bone mask is calculated using reconstructions at two scales which facilitates the distinction of bone and noise voxels. The method was evaluated using scans of a phantom and an observer study. Two observers compared the quality of bone removal with the original and the multi-scale MMBE in CTA-scans of 8 patients. A 4-slice CT-scanner was used (Philips Mx8000, 120 kV, 65/250 mAs (pre-contrast/CTA-scan), 4x1 mm collimation, pitch 0.875).

RESULTS

The phantom study demonstrated that the mean distance of the mask and the bone edge was reduced from 1.0 mm (original MMBE) to 0.5 mm (multi-scale MMBE), with almost the same bone removal effectiveness. The observers had a clear preference for the quality of the bone removal using multi-scale MMBE. In the z-direction sometimes small bone remnants would remain. This will be reduced when scans with a narrower collimation are used.

CONCLUSION

The multi-scale approach results in an improved quality of bone elimination in CTA-images.

DISCLOSURE

M.V.: Financially supported by Phillips Medical Systems(Best) Inc.

Cite This Abstract

Gratama van Andel, H, Venema, H, Streekstra, G, Van Straten, M, Den Heeten, G, Grimbergen, C, Multi-scale Matched Mask Bone Elimination (multi-scale MMBE): Improved Automatic Bone Removal in Computed Tomography Angiography (CTA) Images.  Radiological Society of North America 2005 Scientific Assembly and Annual Meeting, November 27 - December 2, 2005 ,Chicago IL. http://archive.rsna.org/2005/4410663.html