1	Fully automated bone removal and MIP creation of CTA datasets using Matched Mask Bone Elimination (MMBE) and Grid Computing CB Majoie, HW Venema, JG Snel, H Gratama van Andel, M van Straten, CA Grimbergen, GJ den Heeten Depts of Radiology and Medical Physics, Academic Medical Center, Amsterdam, The Netherlands (c.b.majoie@amc.uva.nl)
2	Introduction CT Angiography (CTA) is an important imaging modality for the visualization of intracranial arteries and veins, and cervical arteries. Source images from CTA scans are often displayed as maximum intensity projections (MIPs) to obtain angiographic images. A drawback of the MIP technique is overprojection of bone tissue. Manual removal of the bone from source images (editing) is time consuming. Matched Mask Bone Elimination (MMBE) is a valuable technique to automatically eliminate bone pixels from the source images, to create bone-free MIPs. We have developed and extensively tested a MMBE unit that enables fully automatic, operator independent bone removal and MIP creation using GRID computing. This unit can be integrated in any hospital environment.
3	Purpose To demonstrate our fully automatic bone elimination method, illustrated by several case examples, including intracranial aneurysms, intracranial vascular stenosis, dural sinus thrombosis and cervical artery stenosis.
4	Problem with MIPs from conventional CTA data sets Vessels in the region of the skull base are obscured by bone Extensive (manual) editing is time consuming, and not all bone can be removed
5	MMBE: Technique (ref. 1) Two scans are made: a low dose nonenhanced CT scan and a regular dose CTA scan. The nonenhanced CT-scan and the CTA scan are matched to compensate for patient movements. Bone voxels are identified in the matched nonenhanced CT scan and used as a mask to remove the bone tissue from the CTA scan. A piecewise matching technique is used in CTA scans of the neck, where multiple bones are present (ref. 2) Bone-free MIPs (or Volume Rendered images) are obtained from the masked CTA scans.
6
7
8	Interlude on subtraction and noise MMBE is not a subtraction technique Drawback of subtraction: more noise, reduced image quality
9	Automatic Bone Removal Workflow Obtain scout view for CT acquisition planning Acquire a low dose, nonenhanced CT scan Determine delay time with a test bolus (optional) and administer the contrast agent Acquire CTA scan Nonenhanced CT and CTA scan are sent to the MMBE Grid computer MMBE procedure is performed MIPs (or VR images) are created Images are forwarded to diagnostic viewing station Steps 5 – 8 are performed fully automatic on the Grid.
10	Grid Computing We have developed a computer framework that supports various medical image analyses including MMBE. The Grid computer balances the load of possibly simultaneous MMBE requests (and other analyses) from different CT scanners in our hospital. On receipt of the images using the DICOM protocol the framework schedules the MMBE procedure and MIP creation on a computational grid. After processing, the results are cached and forwarded to a workstation for review and further analysis or to an image archive (PACS).
11	Overview of the Radiology network including the MMBE Grid computer
12	CTA Circle of Willis 57-y-old man with SAH
13	CTA Circle of Willis MIPs of CTA after MMBE
14	CTA Circle of Willis 67-y-old woman with SAH
15	CTA Circle of Willis MIPs of CTA after MMBE
16	CTA Circle of Willis 43-y-old woman with SAH
17	CTA Circle of Willis MIP of CTA after MMBE
18	Intracranial venous system 69-y-o man with clinical suspicion of dural sinus thrombosis
19	MIP of CTV after MMBE
20	Intracranial Venous System
21	MIP of CTV Dural Sinus Thrombosis
22	Cervical Arteries
23	Cervical Arteries
24	Carotid Stenosis
25	Carotid Stenosis
26	Conclusion Fully automated bone removal of CTA data sets is possible using Matched Mask Bone Elimination. The procedure is fast and user transparent with the aid of Grid computing. This technique is valuable for all neuro-CTA applications including CTA of the intracranial arteries and veins, and the cervical arteries. Correspondence: c.b.majoie@amc.uva.nl
27	References Venema HW, Hulsmans FJH, den Heeten GJ. CT angiography of the circle of Willis and the intracranial carotid arteries: maximum intensity projection with matched mask bone elimination. Radiology 2001; 218:893-898 Van Straten M, Venema HW, Streekstra GJ, Majoie CB, den Heeten GJ, Grimbergen GA. Removal of bone in CT angiography of the cervical arteries by piecewise matched mask bone elimination. Med Phys 2004; 31:2194-2933 Majoie CBLM, Van Straten M, Venema HW, den Heeten GJ. Multislice CT venography of the dural sinuses and cerebral veins using matched mask bone elimination. AJNR Am J Neuroradiol 2004;25: 787-791
28	Disclosure The authors have no financial interest or other relationships with any manufacturer of any commercial products or services discussed in this exhibit.