NCCPM

Guildford, United Kingdom
Guildford, United Kingdom
SEARCH FILTERS
Time filter
Source Type

Strudley C.J.,NCCPM | Young K.C.,NCCPM | Oduko J.M.,NCCPM | Looney P.,NCCPM | And 2 more authors.
Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) | Year: 2012

Seven Hologic Dimensions digital breast tomosynthesis (DBT) systems have been installed for the TOMMY trial, a UK based multi-centre trial comparing conventional 2D digital mammography with DBT. In the absence of established guidelines for DBT quality control, a specific protocol was developed and applied. Physics tests of 2D and DBT performance were conducted at baseline and are repeated every 6 months, and include dose to the European standard breast model, tomosynthesis contrast to noise ratio, geometric distortion, z-resolution and threshold contrast detail detection. In addition routine performance checks (daily, weekly and monthly) are conducted on each system and reviewed centrally. Doses delivered under automatic exposure control by the systems were found to be well matched, with a mean glandular dose for the standard breast model (53mm equivalent breast thickness) of 1.89 mGy (range 1.79 to 2.00 mGy) for DBT and 1.41 mGy (range 1.36 to 1.48 mGy) for 2D imaging. Detector performance and image quality measurements were also well matched. All the systems exceeded the achievable image quality standard in the European guidelines for conventional digital mammography. © 2012 Springer-Verlag Berlin Heidelberg.


Oduko J.M.,NCCPM | Young K.C.,NCCPM
Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) | Year: 2012

Contrast-detail measurements were made at approximately weekly intervals for three months, for two full-field mammography systems with different types of detector. The measured threshold contrast values were found to be reasonably stable but with some random variation. The coefficient of variance was 8-10% for detail sizes 0.1 and 1.0mm, and 3-5% for detail sizes 0.25 and 0.5mm. The output of both X-ray sets was also monitored, and found to vary within ±1% of the mean. The variation in threshold contrast is likely to be mainly due to variation of noise in the CDMAM images. Care should be taken when setting baselines and acceptable limits, so that measured changes in threshold contrast that are of the order of ±10% of the mean are not wrongly interpreted as significant changes in performance of a digital mammography system. © 2012 Springer-Verlag Berlin Heidelberg.

Loading NCCPM collaborators
Loading NCCPM collaborators