Computer-aided detection of mammograms improves breast cancer screening

18 October 2005

Aberdeen, UK and Bedford, Mass, USA. In screening mammograms for breast cancer, the performance of radiologists using computer aided detection (CAD) was superior to double reading, according to a recently published UK study.

The study titled "Single reading with computer-aided detection and double reading of screening mammograms in the United Kingdom National Breast Screening Program" was published in the October issue of the journal Radiology.*

This landmark study by researchers from two leading breast centres in the United Kingdom measured radiologists' performance with the use of CAD. The study also showed a 15% increase in cancer detection rates when CAD was used. The CAD system (ImageChecker 1000 version 5.0) used in this study was supplied by R2 Technology, a subsidiary  of diagnostic imaging supplier Hologic Inc. (NASDAQ:HOLX).

The CAD program scans a mammogram for suspicious features or abnormalities that could represent breast cancer. When the computer finds anything unusual it indicates the abnormality on a screen for the radiologist to check. The CAD system used in this study features proprietary R2 EmphaSize(TM) software, which provides variable size marks that correlate to lesion significance. A larger mark indicates that the CAD algorithm detects more mammographic features that are indicative of cancer. The marks prompt the radiologist to double check suspicious areas, particularly focusing on features with larger CAD marks.

The study included a sample of more than 10,000 mammograms obtained from women aged 50 years or older who underwent routine screening. Mammograms that were double read initially were randomly allocated to be re-read by eight different radiologists using a single read and CAD. Cancer detection rates and recall rates from double reading and single reading with CAD were then compared. The researchers found that single reading with R2 CAD detected 6.5% more cancers — a 15% increase in the cancer detection rate — than that achieved with double reading (P=0.02)

Professor Fiona Gilbert, of Aberdeen University, who led the study, said: "The results of this trial are very encouraging. The mammograms studied were from a sample taken in 1996 so that all cancers that developed subsequently in this group of women could be included. The study was retrospective so the radiologists taking part in the trial knew no action would be taken as a result of their decisions."

"We have now embarked on a new study to confirm that the CAD result is still as good when used in real day to day decision making about breast cancer diagnosis," Professor Gilbert added.

Dr. Sue Astley of Manchester University who also worked on the study said: "This new prospective trial will involve 30,000 women in three major screening centres, most of whom will have single reading with CAD in addition to their routine double reading. This is an opportunity for women in the UK to have their mammograms read using the latest CAD technology which is already available in America and other countries."

To top

To top