Evaluation of absorbed dose and image quality in mammography

by Hemdal, Bengt, PhD

Abstract (Summary)
Mammography refers to the X-ray examination of the human breast, and is considered the single most important diagnostic tool in the early detection of breast cancer, which is by far the most common cancer among women. There is good evidence from clinical trials, that mammographic screening can reduce the breast cancer mortality with about 30%. The side effects include a small and age related risk of carcinogenesis due to the exposure of the glandular tissues in the breast to ionising radiation. As for all X-ray examinations, and of special importance when investigating large populations of asymptomatic women, the relationship between radiation risk and diagnostic accuracy in mammography must be optimised. The overall objective of this thesis was to investigate and improve methods for average glandular dose (AGD) and image quality evaluation in mammography and provide some practical guidance. Language of Abstract: English Swedish Arabic Chinese Danish Finnish French German Greek Hebrew Icelandic Italian Japanese Latin Norwegian Polish Portuguese Romanian Russian SpDose protocols used for so-called reference dose levels in Sweden 1989 (Nordic) and 1998 (European) were compared in a survey of 32 mammography units. The study showed that the AGD values for a “standard breast” became 5±2% (total variation 0?9%) higher at clinical settings, when estimated according to the European protocol. For the Sectra MDM, a digital mammography (DM) unit with a scanning geometry, it was impossible to follow procedures for characterisation of the X-ray beam (HVL=half value layer) specified in the European protocol. In an experimental setup, it was shown that non-invasive measurements of HVL can be performed accurately with a sensitive and well collimated semiconductor detector with simultaneous correction for the energy dependence. AGD values could then be estimated according to 3 different dose protocols. A dosimetry system based on radioluminescence and optically stimulated luminescence from Al2O3:C crystals was developed and tested for in vivo absorbed dose measurements. It was shown that both entrance and exit doses could be measured and that the dosemeters did not disturb the reading of the mammograms. A Monte Carlo study showed that the energy dependence could be reduced, primarily by reducing the diameter of the crystal. It is proposed that radiation scattered forward towards the breast from the compression paddle, a scanning device etc, should be considered with greater clarity in the breast dosimetry protocols, and be described with a forward-scatter factor, FSF, for the various geometries and conditions proposed. Low contrast-detail (CD) phantoms of simulated glandularity 30, 50 or 70%, and thickness 3, 5 or 7 cm, were used to compare three different mammography systems. The same number of perceivable objects was visible for the full-field DM system at 20-60% of the AGD necessary for the screen-film (SFM) system, with the largest dose reduction potential for the thickest phantoms with the highest glandularity. However, more recent research shows that CD phantoms with a homogeneous background, as used here, must be used with care due to the presence of “anatomical noise” in the real clinical situation. Image quality criteria (IQC) recommended in a European Guideline 1996 for SFM were adjusted to be relevant also for DM images. The new set of IQC was tested in two different studies using clinical images from DM and SFM, respectively. The results indicate that the new set of IQC has a higher discriminative power than the old set. The results also suggest that AGD for the DM system used may be reduced.
Bibliographical Information:


School:Lunds universitet

School Location:Sweden

Source Type:Doctoral Dissertation

Keywords:MEDICINE; NATURAL SCIENCES; Physics; radioluminescence; optically stimulated luminescence; in vivo dosimetry; dose reduction; dose protocol; average glandular dose; absorbed dose; Breast cancer; mammography; Monte Carlo; image quality; image quality criteria; contrast-detail phantom


Date of Publication:01/01/2009

© 2009 All Rights Reserved.