Assessment of mammographic CAD
Research type
Research Study
Full title
Assessment of the sensitivity and specificity of the VuCOMP computed aided detection (CAD) system for breast cancer detection on digital mammography images
IRAS ID
132662
Contact name
Anthony J Maxwell
Contact email
Sponsor organisation
University Hospital of South Manchester NHS Foundation Trust
Research summary
Mammography screening programmes reduce mortality by early detection of breast cancer. The mammographic signs are often subtle. Double reading of the images by two readers increases the detection rate compared to single reading.
Computer aided detection (CAD) is designed to act as an additional reader. CAD works by applying computer processing to the digital images. Abnormalities such as masses or areas of microcalcification are marked by ‘prompts’ such as circles or asterisks.
CAD has a similar sensitivity for cancer detection to human readers but lacks specificity, with >1.5 false prompts per case on average. The reader may ignore correct prompts for cancers and act on false positive prompts. VuCOMP Inc. claims that its CAD has superior performance to its competitors, with a higher detection rate and fewer false positive prompts.
The aim of the study is to assess the performance of the VuCOMP CAD system on a retrospective series of normal and abnormal mammograms and to determine whether breast density affects CAD performance. The results will be used to inform the design of further studies. If ultimately it can be shown that the performance of one human reader plus CAD is equivalent to two human readers then this would reduce the manpower requirements for the breast screening programme and ensure its sustainability.
REC name
East of England - Cambridgeshire and Hertfordshire Research Ethics Committee
REC reference
13/EE/0188
Date of REC Opinion
10 Jun 2013
REC opinion
Favourable Opinion