Computer-Aided Detection is Effective

Picture: A breast cancer cell

Scientists at The Universities of Manchester and Aberdeen and Cancer Research UK have found that the workload of radiologists could potentially be halved by using a new computer-aided system to help read breast x-rays and detect cancer.

With the computer-aided detection (CAD) system, only one expert is needed to look at each mammogram, rather than the usual two. Use of the system could free up the experts' time, enabling more face-to-face consultations and all women to be screened as often as recommended. The CAD programme searches mammograms for suspicious features or irregularities that could be caused by cancer. When the computer finds anything unusual it indicates it on a screen for the radiologist to look at.

Study results published today show that mammogram readings by a single expert plus the CAD system may be as good as those read by two expert radiologists, and in some cases the new combination could be even more successful.

The researchers took more than 10 000 mammograms that had previously each been read by two radiologists. These were then read again by a single radiologist, who was prompted by the computer to double-check suspicious areas for any abnormalities. The results showed that the cancer detection rate by a single reader using CAD was at least as good as that when the films were originally read by two readers.

The mammograms studied were from 1996, so that all cancers that developed subsequently in the group of women could be included, and no action was taken as a result of the radiologists' decisions.

Dr Caroline Boggis, Consultant Radiologist at the Nightingale Breast Centre in Manchester said: "The results of this first trial are very encouraging, and we have just started a new study to confirm that the results of using CAD are still as good when used in real decision-making in the breast screening programme."; Source: University of Manchester