Battle Strategies: Winning the Multidetector CT Data War

A heart imaged in real-time 3D on the GE Lightspeed VCT 64-slice scanner.

Sixty-four slice CT is a tremendous clinical boon for cardiac imaging, opening the door to non-invasive images of the heart, enabling physicians to diagnose cardiac disease at earlier stages and providing the data needed to allow some patients to bypass invasive cardiac procedures. Despite its clinical advantages, 64-slice CT is not without challenges. Managing the image volume beast sits near the top of the list.

Sixty-four slice systems churn out massive amounts of data with scanners regular producing 4,000 images or up to 20 gigabytes of raw data per scan. The data load is enough to tax even the most meatiest storage systems, which forces sites to wrangle with some tough questions, including:

• Which images should be saved on PACS?
• What other media can be used to store multidetector CT data?
• What is the best way to balance the regulatory and economic needs of the practice or facility?
• What is the ideal system to move and share images?

This month, a few cutting-edge sites outline their 64-slice CT data management plans to help steer their colleagues toward an optimal solution.


Click and burn

“The New York Presbyterian Hospital [New York City] is ultra-meticulous,” confirms James Min, MD, director, cardiac CT laboratory at Weill Cornell Medical College of The New York Presbyterian Hospital. New York Presbyterian employs a three-pronged approach toward storage of images generated by its GE Healthcare VCT XT scanner. The practice deployed 64-slice cardiac imaging in August 2005 and completes 70 to 100 studies monthly. Typical studies include retrospective gating and functional motion analysis and generate 1,000 to 4,000 images each.

Although the 64-slice scanner produces a hefty amount of data, only a fraction of the original images are stored on PACS. Images designated for PACS storage include a single axial dataset of approximately 200 images, scout images and calcium scores, which comprise between two and 40 images. The next layer of data includes reconstructions for functional evaluation; typically cardiologists review 20 phases or axial datasets. These images are stored on PACS and backed up onto optical disk. Finally, the hospital adds an extra layer of security to its storage plan. To meet the complementary goals of answering questions—such as left ventricular ejection fraction—at later dates and ensuring complete regulatory compliance and maintaining all reviewed studies for seven years, the practice burns the entire exam to a DVD.

“Each DVD costs 30 to 40 cents; this approach is probably more cost-effective than PACS storage for small practices,” opines Min. Still, the DVD option is not without its downsides. “It does take longer and can use more staff resources than outright PACS storage,” states Min, who admits the practice could shave time off the process by requiring techs to burn and label studies at the time of acquisition.


Triple redundancy

CVCTA Education in San Francisco employs a triple redundant approach to handling data generated by its Toshiba America Medical Systems Aquilion 64-slice CT scanner. The cardiac CT education center performs about 15 to 20 scans daily including eight cardiac scans and three to four peripheral vascular studies.

A typical cardiac scan produces about 20 gigabytes of raw data, an amount that would overwhelm most storage solutions. The center completes approximately 15 reconstructions on cardiac scans or 1.2 gigabytes of data. “We save functional data, one or two thin-slice reconstructions and 10 thick-slice reconstructions for coronary studies on PACS. We wind up archiving about 750 megabytes of data per study,” says Tony DeFrance, MD, medical director. Techs also burn a DVD with the same 750 megabytes of data directly from the scanner. “Patients and doctors appreciate the advantages of access to a hard copy,” reports DeFrance. The final layer of CVCTA’s approach is driven by its teaching operations and consists of a 10 terabyte server that allows the site to easily distribute studies to multiple workstations as CVCTA trains physicians new to cardiac CT.

The other component of 64-slice scanning is image review. Sites need to determine who is reading studies and where. “Sites need good bandwidth, usually T1 or T2 lines to move studies around,” says DeFrance. 


Seamless storage

But 64-slice CT need not be a storage beast. Take for example South Jersey Radiology Associates, a seven-site practice headquartered in Voorhees, N.J., that operates the CT continuum from four-slice systems to Siemens Medical Solutions Sensation 64 and Definition Dual Source scanners. The practice runs a full schedule on its scanners and completes 40 to 45 CTs daily on each of its mega scanners: the Definition and Sensation 64. Both scanners yield datasets of similar sizes. A typical chest CT from the Definition produces 206 images and consumes 37 megabytes of PACS storage; a chest CT from the 64-slice scanner generates 208 images, which requires 39 megabytes of storage space. Despite the file sizes, multidetector CT has been fairly seamless from a storage perspective, says PACS Administrator Kim Baldwin.

The practice does employ a triage approach to data storage. For standard exams like chest, abdomen and pelvis studies, South Jersey Radiology sends 5 millimeter (mm) axial reconstructions and 3 mm coronal reconstructions to PACS. Studies requiring 3D manipulations such as coronary CT angiographies require a different approach. “We save the thin slice [data] on Siemens syngo WebSpace. Radiologists can interact with the slices without eating up PACS storage,” explains William Muhr, MD, radiologist. Any images created by the radiologists are then saved to PACS. Like most cardiac CT sites, the practice does not hold on to raw data for the long-term, relying on the Siemens Leonardo workstation as a short-term, temporary storage archive.