RSNA: New automated CCTA software may be useful in triaging ED patients
CHICAGO—Negative results from an automated coronary CT angiography (CCTA) reader may be useful for triaging emergency department (ED) patients, based on a study presented Monday at the Radiological Society of North America (RSNA) conference. Conversely, the researchers determined that positive automated results require further interpretation by an experienced reader.
Principal investigator and study presenter Girish Tyagi, MD, from the department of radiology at Beth Israel Deaconess Medical Center in Boston, and colleagues assessed the accuracy of COR Analyzer (Rcadia Medical Imaging in Haifa, Israel), an automated software analyzer of CCTA.
The researchers retrospectively studied 21 CCTAs performed on ED patients with chest pain and low to moderate probability of suspected coronary artery disease (CAD) between September 2008 and February 2009 using the COR Analyzer, which is FDA approved.
In the 100 analyzable studies, they compared the automated results for 10 coronary artery segments to the PACS interpretation by an expert CCTA reader for a total of 210 segments. Tyagi noted that seven patients’ images were excluded due to “technical glitches,” and four patients were excluded because the image quality was not “up to standard.”
Utilizing the interpretation of an expert CCTA reader as gold standard, Tyagi reported that the automated results from COR Analyzer for significant (greater than 50 percent) stenosis in 210 coronary artery segments yielded three true positives, 15 false positives, 192 true negatives and one false negative.
Overall, the software identified five of six patients determined by the experts to have significant stenosis. On a per patient basis, there were “five true positives, 16 false positives and one false negative, leading to an overall sensitivity of 83 percent, specificity of 82 percent and a negative predictive value of 92 percent.”
Overall, the automated CCTA reading in the study had a sensitivity of 75 percent, specificity of 93 percent, a positive predictive value of 17 percent and a negative predictive value of 99 percent.
Based on their findings, Tyagi concluded that “computer-aided detection systems may have a role, especially in the ED setting as an aid to CCTA interpretation.”
He also added that there are some groups that are beginning to utilize the technique in clinical practice.
Principal investigator and study presenter Girish Tyagi, MD, from the department of radiology at Beth Israel Deaconess Medical Center in Boston, and colleagues assessed the accuracy of COR Analyzer (Rcadia Medical Imaging in Haifa, Israel), an automated software analyzer of CCTA.
The researchers retrospectively studied 21 CCTAs performed on ED patients with chest pain and low to moderate probability of suspected coronary artery disease (CAD) between September 2008 and February 2009 using the COR Analyzer, which is FDA approved.
In the 100 analyzable studies, they compared the automated results for 10 coronary artery segments to the PACS interpretation by an expert CCTA reader for a total of 210 segments. Tyagi noted that seven patients’ images were excluded due to “technical glitches,” and four patients were excluded because the image quality was not “up to standard.”
Utilizing the interpretation of an expert CCTA reader as gold standard, Tyagi reported that the automated results from COR Analyzer for significant (greater than 50 percent) stenosis in 210 coronary artery segments yielded three true positives, 15 false positives, 192 true negatives and one false negative.
Overall, the software identified five of six patients determined by the experts to have significant stenosis. On a per patient basis, there were “five true positives, 16 false positives and one false negative, leading to an overall sensitivity of 83 percent, specificity of 82 percent and a negative predictive value of 92 percent.”
Overall, the automated CCTA reading in the study had a sensitivity of 75 percent, specificity of 93 percent, a positive predictive value of 17 percent and a negative predictive value of 99 percent.
Based on their findings, Tyagi concluded that “computer-aided detection systems may have a role, especially in the ED setting as an aid to CCTA interpretation.”
He also added that there are some groups that are beginning to utilize the technique in clinical practice.