CHICAGO, Nov. 29—Computer-aided detection (CAD) is effective for the localization of the often difficult to detect morphologically flat early colonic cancerous lesions, as well as later stage colorectal cancers, according to two clinical trials presented at the 93rd annual meeting of the Radiological Society of North America (RSNA) this week.
Stuart Taylor, MD, radiologist at the University College Hospital, London, “These results are interesting because CT colonography (CTC) is well established for the detection of adenomatous polyps in asymptomatic patients and the role of CAD is becoming increasingly established, but the potential for computer-aided cancer detection in symptomatic patients has been relatively neglected.” The new data support CAD’s use in symptomatic patient populations.
One of the studies involved 30 symptomatic patients who underwent a diagnostic colonoscopy and were undergoing cancer staging. Their tumors were all endoscopically classified as having a flat morphology and were located and characterized by three independent radiologists in conjunction with the endoscopic reports and imaging. CAD software was then applied at three different settings of sphericity. The analysis showed CAD detected 83.3 percent of the morphologically flat cancers present.
Previously, this type of lesion has historically been difficult to detect with CTC, according to Taylor.
The second larger study included 59 symptomatic patients with already proven cancers. CAD was used at four different filter settings and results based on the sensitivity and specificity at the different settings showed CAD to be effective for the detection of these later stage cancers.
Also, results showed that as in the detection of polyps, optimal results for CAD detection of the cancers requires scans to be performed in both supine and prone CT acquisitions, according to Taylor.
Medicsight ColonCAD software was used in both trials.