CAD proves effective when used in CR mammography

May 11, 2007

Data presented at the American Roentgen Ray Society meeting data this week show that computer-aided detection does an effective job of finding breast cancer in mammograms made using computed radiography. The results, obtained by radiologists at George Washington University, will help lay the groundwork for the adoption of CAD for this application, according to iCAD, a pioneer of this technology.

Data presented at the American Roentgen Ray Society meeting data this week show that computer-aided detection does an effective job of finding breast cancer in mammograms made using computed radiography. The results, obtained by radiologists at George Washington University, will help lay the groundwork for the adoption of CAD for this application, according to iCAD, a pioneer of this technology.

The Nashua, NH-based company is awaiting an FDA decision on its submission to market a specific algorithm for use on a CR mammography system by Fujifilm Medical System USA that the agency approved for marketing late last year. Many of the 9000 mammography facilities in the U.S. screen relatively low numbers of women for breast cancer, according to Ken Ferry, iCAD chief executive officer, making them ideal candidates for CR mammography, an inexpensive alternative to full field digital systems that use silicon- or selenium-based detectors.

"There is no doubt that CR will be an increasingly important technology in breast imaging," said Dr. Rachel Brem, director of breast imaging and intervention and vice chair of radiology at the George Washington University Medical Center. "Multiple studies have unequivocally demonstrated the improved detection of breast cancer with CAD. Therefore, we wanted to investigate the synergies of these two technologies."

Results presented at the ARRS meeting were based on CR images made from 53 breast cancer cases that were evaluated using iCAD software. The researchers assessed the sensitivity of cancer detection by CAD as well as the mammmographic density and size of cancers found. Among the 47 of 53 cancer cases that were detected by CAD, 30 were in nondense breasts and 17 were in dense breasts. CAD picked up 11 of 12 cancers manifesting as calcifications and 36 of 41 masses.

"CAD had a high sensitivity of 89% with CR mammography that was maintained even in conditions that may lower the sensitivity of mammography, such as dense breasts and small lesions 1 mm or less," Brem said.

This level of performance is only the beginning, according to Ferry. He expects the sensitivity of the technology to rise as company engineers become more experienced with the Fuji CR system and refine their algorithm. This experience will come, however, after the iCAD technology gets into mainstream use.

The FDA has been reviewing the iCAD software for CR since last July, a process lengthened by an unprecedented level of scrutiny, according to Ferry. He summarized for DI SCAN the present regulatory situation as "different reviewers, different expectations," compared with previous CAD algorithms developed for digital detector systems.

Ferry and his staff met with FDA officials April 20 to discuss issues involved in the review and were promised an update by mid-May.

If iCAD can get through what Ferry describes as this "knothole" in the review process, the company can begin selling its product for CR mammography, an opportunity that he said could be "a huge revenue bonus."

This bonus, if realized through the company's alliance with Fuji, could be just the start. Last November, Caresteam Health (formerly Kodak) submitted the final module of its premarket approval submission for CR mammography. If the Carestream system passes regulatory review, iCAD will submit a new application to the FDA for its algorithm to work with it.