The Next Steps in Cardiac Advanced Visualization

Cardiac computer-aided detection and putting 3-D Cardiac computer-aided detection and putting 3-D
By: 
Dave Fornell

 

March 18, 2011
The University of Southern California is developing an algorithm to produce animated MRI scans of anatomy in motion.

Two big advances in cardiac advanced visualization software were highlighted in December at the 2010 Radiological Society of North America (RSNA) meeting in Chicago. TeraRecon highlighted a computer-aided detection (CAD) software module to detect coronary artery stenosis. Ziosoft showed the latest version of its PhyZiodynamic software, which creates a video loop of a 3-D beating heart from a computed tomography (CT) or magnetic resonance imaging (MRI) dataset. Both of these advances were shown as works-in-progress at RSNA, but are expected to be commercially released in 2011.
Two other advances in advanced visualization at RSNA included software now being accessible on via the Internet on thin-client computers, and compatibility with the iPad.

3-D Hearts in Motion
Advanced visualization can make great pictures of 3-D anatomy, but in the case of the heart, it has so far failed to show functionality and natural movement during the cardiac cycle. Picture archiving and communication system (PACS) users can flip through sets of stacked images from a computed tomography (CT) or magnetic resonance imaging (MRI) dataset to make a sort of primitive cartoon, but it lacks the smooth flow of real anatomical motion.

Using supercomputing algorithms previously used to forecast weather, Ziosoft developed the PhyZiodynamic software to take a CT or MRI data set and put it into motion. At its RSNA booth, Ziosoft offered a full 3-D rendering of a heart in motion (5-D) and invited attendees to rotate and virtually dissect it slice-by-slice while it continued to beat. This offered views of the myocardium and valves in motion as only previously seen by surgeons during open-heart procedures. The views also showed the movement of the valve leaflets and the chordae tendineae. The user could also use the mouse to navigate in the 3-D lumens of the coronary vessels.

The software also has a velocity mode feature, which uses a color-coded overlay to show the stress/motion of the tissue. Researchers at Massachusetts General Hospital in Boston are using this velocity mapping feature in a study to determine viable tissue for the optimal placement of the third lead from a unique defibrillator for very sick patients.

The main application of this technology is in structural heart surgical and transcatheter procedure planning, including transcatheter valve replacements and septal shunt occlusion. It may also allow a new breed of hybrid imaging in the cath lab, where live angiography X-ray could be synchronized with a 3-D cardiac reconstruction in motion. This would enable a real-time synchronization of the images with the motion of the beating heart rather than a fixed/frozen reference image, as is the current hybrid imaging standard.

PhyZiodynamic processing of CT datasets is giving new insights into cardiac dynamics and valve function, said John A. Rumberger, M.D., Ph.D., FACC, director of cardiac imaging, The Princeton Longevity Center, Princeton, N.J. Standard advanced visualization software allows users to flip through a stacked deck of images in a sequence to create a crude representation of motion. However, Rumberger said there is data missing from image to image, so the movement is unnatural and incomplete. He said it is also difficult to assess when a heart valve is at its highest and lowest peaks to accurately assess valve function.

“The traditional reconstruction of the valves is okay, but not great,” he explained. “But Ziosoft’s system is great. It helps fill in the blanks between, voxel-by-voxel, for a smooth transition. It connects the dots. It’s really an incredible step – it’s an amazing technology.”

He showed the new technology to cardiac surgeons who are performing transcatheter valve implants. They said they believe this will be a new way to examine valve function, plan procedures and review post-procedure success. “I don’t think the standard reconstruction of the aortic root is sufficient to make these evaluations,” Rumberger said.

Another application for the PhyZiodynamic software might be to evaluate coronary artery plaque and perfusion in the myocardium. He explained the system eliminates gaps in data from frame-to-frame and its noise filtering algorithms might overcome the hurdles that prevent total CTA analysis. The technology might be able to cut the number of diagnostic catheterization angiograms.

“If I can start looking at the dynamics of plaque in the coronary vessels, that is the holy grail,” Rumberger said. “If I can use a CT image to look at perfusion, I don’t need anything else.”

Years ago, Rumberger worked on experiments using electron beam CT to evaluate cardiac perfusion, but beam hardening caused issues with noise and image artifacts. He said PhyZiodynamic appears to correct these beam-hardening artifacts. Iterative reconstruction software offered by many CT vendors as a way to lower radiation dose helps filter much of the noise on CT scans. PhyZiodynamic works in a similar fashion, but can clean up image quality on noisy CT datasets that are months or even years old, Rumberger said. “That’s an incredible advantage,” he said. It might also be possible to use the software to help make better reconstructions of lower-dose CT scans moving forward.

Supercomputing has overcome the first major obstacle to this technology by enabling massive amounts of computing power to reduce the dynamic image processing time. At the American College of Cardiology (ACC) meeting in March 2010, the PhyZiodynamic software took about eight hours to process a cardiac cycle dataset, which limited its practical application. However, by RSNA 2010, the company cut the processing down to about 45 minutes and is working to reduce it further.

The second obstacle that needed to be overcome to make this technology viable was new filtering technology to remove noise, motion and stitching artifacts from the video loop. The first version of the software previewed at ACC filtered about 80 percent of the blood/noise artifacts in the ventricles and vessels. By RSNA the software was able to filter about 99 percent of the noise artifacts.

While PhyZiodynamic works well with high 256- and 320-slice scanners, Ziosoft designed the software so it’s compatible with the standard 64-slice CT systems most hospitals use.

The company hopes to have a commercialized version of the software available in 2011.

Heart CAD is Close to Release
Computer-aided detection (CAD) software algorithms are currently used to help spot potential areas of interest to a radiologist, such as possible breast lesions in mammograms or early lung cancer. The system acts as a second set of eyes to read imaging exams. TeraRecon is about to introduce the first CAD system to help detect coronary artery stenosis from CT or MRI datasets.

The company displayed its Centerline Crossline Analytics plaque analysis software at RSNA. It automatically analyzes the datasets to find the coronary vessels. It then pinpoints areas of interest, such as severe coronary stenosis, with a dot. TeraRecon President and CEO Robert Taylor, Ph.D., hopes the new feature will be released in the middle of 2011.

Cloud-Computing, iPad Functionality
In the past couple of years, cloud-computing has been introduced, allowing the computing power of a large server to be accessed by smaller, thin-client computers via the Internet. This allows computing-power hungry software, such as advanced visualization, to be accessed and used by physicians on their PCs from any location without having to load the software on their own computer. Vital Images, Shina Systems, TeraRecon and Ziosoft all offer cloud/Web-based applications.

Rumberger uses both Ziosoft and TeraRecon’s advanced visualization cloud-computing applications. He is amazed at the computing power he can now access. “I can do all the advanced visualization on my laptop as opposed to using my desktop workstation,” he said.

“We believe that the future of medical imaging will inevitably consist of a combination of on-site software, cloud-based hosting and meaningful integration with other enterprise healthcare informatics systems,” Taylor said when TeraRecon launched its iNtuition Cloud software last year.

Cloud-based computing also enables small computers to manipulate large datasets. One of the biggest trends at RSNA 2010 was advanced visualization and PACS software integration with iPad devices so medical images and reports can be accessed and edited anywhere. This introduces a new concept of what constitutes a workstation and unchains the radiologist or cardiologist from their desktop, thick-client computer. Many vendors also believe being able to call up images anywhere will enhance collaboration between physicians in different hospital departments, referring physicians and patients.

In February, the U.S. Food and Drug Administration (FDA) cleared Mobile MIM, the first mobile radiology application that allows physicians to view and make a diagnosis from medical images on the Apple iPhone and iPad. The FDA said the application is not intended to replace full workstations and is indicated for use only when there is no access to a workstation.

However, the application’s clearance will likely be the first of many, as physicians are rapidly adopting iPads as a replacement for traditional clipboards. The device offers a mobile workstation to access and enter information for patient records, images and test results. RSNA 2010 event was unofficially dubbed “the year of the iPad” by many due to the vast number of vendors showing new iPad applications to access and manipulate radiology images.

Sidebar

MRI 5-D Imaging in Development Researchers at the University of Southern California announced in February they developed an algorithm to produce animated magnetic resonance imaging (MRI) scans. It can reconstruct an animated representation of any area of a living person. Reconstructed skeletal joints can be moved with a 3-D animation program, and the soft tissues deform interactively and realistically according to the MRI scan data. The underlying approach is an intelligent, example-based interpolation technique called volume blend deformation. The challenge in applying this technique is accurate registration of scans in different poses. This required the development of a new hierarchical skeleton-guided registration step. Although the registration requires many hours of pre-computation, the final result can be animated interactively. The new algorithm will be particularly valuable in visualizing problems related to movement, since it visualizes the internal structures in motion. The work was done in conjunction with the Samsung Advanced Institute of Technology and researchers from Weta Digital and Victoria University in Wellington, New Zealand. It will be published in the March 2011 issue of the journal IEEE Transactions on Visualization and Computer Graphics (Taehyun Rhee, J.P. Lewis, Ulrich Neumann, and Krishna Nayak. “Scan-Based Volume Animation Driven by Locally Adaptive Articulated Registrations.” IEEE Transactions on Visualization and Computer Graphics, March 2011, 17, 3, pp. 368-379.)

  • The University of Southern California is developing an algorithm to produce animated MRI scans of anatomy in motion.
The content of this field is kept private and will not be shown publicly.
Type the characters you see in this picture. (verify using audio)
Type the characters you see in the picture above; if you can't read them, submit the form and a new image will be generated. Not case sensitive.