Taylor, P; Wilson, S; Potts, H; Wilkinson, L; Khoo, L; Given-Wilson, R; (2009) Evaluation of CAD with Full Field Digital Mammography in the NHS Breast Screening Programme. (NHSBSP Equipment Report 0910 ). NHS Cancer Screening Programmes: Sheffield.
Full text not available from this repository.
An evaluation of the use of computer aided detection with full field digital mammography was undertaken with the support of the NHS Breast Screening Programme. It included the following components a. a study of the effect of computer aided detection (CAD) on workflow and the time taken to read a mammographic image b. a systematic review and meta-analysis of studies comparing: single reading of images with and without CAD; double and single reading of images without CAD c. an assessment of the cost-effectiveness of CAD in full field digital mammography (FFDM) d. a review of issues relating to conformance with key standards. The main points to emerge under each heading were as follows a. Supplementing FFDM with CAD did not increase reading time. However the time taken to read each screening case was increased substantially by the use of FFDM rather than analogue film. Using CAD with FFDM was found to be simple and intuitive and the majority of readers would use it again. The addition of CAD had no demonstrable effect on the number of women recalled for arbitration or assessment; the number of cases examined was small, however, and only gross effects would have been detected. b. A systematic review and meta-analysis was conducted of studies produced up to 2008 and has been published elsewhere.1 Ten of these studies compared single reading of images with CAD and without it. They examined 410 000 subjects and found that CAD did not significantly increase sensitivity in cancer detection (1.04, 0.96–1.13). Seventeen studies compared single reading (SR) and double reading (DR) of images without CAD. They examined 1.2 million subjects and showed increased cancer detection rates with DR (1.10, 1.06–1.14). Recall rates increased for both CAD+SR (1.10, 1.09–1.12) and DR (1.17, 1.15–1.18) but decreased for DR with arbitration (0.94, 0.92–0.96). A later randomised controlled trial of CAD+SR compared with DR showed that they are equally effective at cancer detection.2 The cost-effectiveness of CAD+SR with FFDM was compared with DR, focusing on the amount of time a radiologist needed to read cases and deal with the women recalled. An earlier study had shown that using CAD prompts with analogue film reading almost doubled reading time and demanded more radiologist time than did DR.3 c. For its analysis of CAD with FFDM the present report considered a range of possible increases in recall rates, in the time allocated for assessment, and in the number of CAD licences purchased. It demonstrated that it is possible to save radiologist time overall and reduce cost by using CAD+SR rather than DR. d. Conformance with relevant standards – Integrating the Healthcare Enterprise (IHE) and Digital Imaging and Communications in Medicine (DICOM) standards – is discussed briefly where it relates to installing CAD with FFDM. In summary, CAD with FFDM was found to be user friendly and did not affect radiologist workflow or reading time during film-screen reading. In the published studies, estimates vary of the extent to which using CAD increases recall rates; depending on these rates, and the time taken to assess a woman, CAD can save radiologist time during the screening and assessment process when compared with DR. Systematic review of published literature comparing CAD+SR with SR showed no significant increase in cancer detection rates and an increase in recall rates. When compared with SR, however, DR with arbitration significantly increased cancer detection and decreased recall rates.
|Title:||Evaluation of CAD with Full Field Digital Mammography in the NHS Breast Screening Programme|
|UCL classification:||UCL > School of Life and Medical Sciences > Faculty of Population Health Sciences > Institute of Epidemiology and Health Care > CHIME|
Archive Staff Only: edit this record