Automated Cardiac-CT Reader Helps ED Sort Out Chest-Pain Patients in Off Hours

Reed Miller

December 01, 2009

December 1, 2009 (Chicago, Illinois) — Automated cardiac CT-analysis software has a high negative predictive value and may help move patients quickly through the emergency department, a small study at Beth Israel Deaconess Medical Center, in Boston, MA, shows [1].

For patients presenting to the emergency department with chest pain, "the algorithm of ECG, serial enzymes, and perfusion imaging is both expensive as well as time-intensive, so clearly there is a need for a quick noninvasive test, and recent studies have shown that [coronary computed-tomography angiography] CCTA fits the bill very nicely. But there are problems with CCTA--one of them is the off-hours availability of expert coronary CTA readers," Dr Girish Tyagi explained here at the Radiological Society of North America 2009 Scientific Assembly.

"The automated analyzer could have a role as a second reader in the emergency-department setting as an aid to normal expert CCTA readers. It's available at all hours, it interprets raw data--there's no waiting for 3D reformats--and there are cost savings to be had from rapid triage of this group of patients," he said.

Tyagi presented data from his center's initial experience with the COR Analyzer (Rcadia Medical Imaging, Auburndale, MA), a software system that reads and displays CCTA images of major coronary artery segments.

In the study, the COR Analyzer processed CCTA images from 115 low- to intermediate-risk patients who came to the hospital with suspected coronary disease. Tyagi and colleagues compared the COR analyses with the consensus interpretation of two expert readers.

For 100 analyzable studies, the automated analysis yielded a negative predictive value of 98%. The COR Analyzer's overall sensitivity was 83% and its specificity was 82%. Tyagi characterized all three figures as "high." The COR Analyzer identified five of the six patients with significant stenosis found by the expert readers (five true positives, one false negative). COR also yielded 16 false positives and 78 true negatives.

Software Aids in Triage

Tyagi observed that the automated CCTA analyzer errs to a high number of false positives, especially in patients with calcified plaques, and that positive results by the COR Analyzer will always have to be followed by further interpretation by an expert reader. "We're not suggesting this should be placed before clinical reading, but as a quick triage, at least the clinicians can start looking for alternate origins of chest pain [if the result is negative]." Despite the limitations, Tyagi believes "the system may provide nonexpert readers with confidence to rule out significant stenosis, leading to decreased length of stay for patients in the emergency department and improved throughput there."

Commenting on the study, Dr Kavitha Chinnaiyan (William Beaumont Hospital, Royal Oak, MI) told heartwire that she believes automated CCTA has "potential use in emergency-department situations, particularly in those centers that don't have readers around the clock." She said that the negative predictive value of the test shown in the trial at Beth Israel Deaconess is "great" and that the results overall were very good.

However, she insists that the technology is not yet "ready for prime time" and that it must be further validated in much larger trials and compared with coronary angiography in order to bring its accuracy closer to that of expert human readers. She noted that the sensitivity of 83%--the COR Analyzer missed one patient with CAD in the trial--is "a little worrisome."

Tyagi and his colleagues are currently developing a larger, longer trial of the COR Analyzer that will either randomize patients to automated analysis vs human analysis only or will analyze the CCTA of every low- to moderate-risk chest-pain patient with COR over a six-month period. He said the launch of that study is currently working its way through the institutional review board process.

Tyagi disclosed having no conflicts of interest.

Comments

3090D553-9492-4563-8681-AD288FA52ACE
Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.

processing....