Materials and Method:
The study has been performed in a Siemens Inspiration unit installed at a University Hospital.
Dose management online system connected to this modality is DoseWatch (General Electric Healthcare).
215 patients undergoing bilateral mammography with two projections: mediolateral oblique (MLO) view and craniocaudal (CC) view have been included.
AGD and ESAK values are provided by the mammographic system for each image as dosimetric indices,
and therefore extracted by the DMP.
Prior to its use for dosimetric purposes,
indices have to be validated by a medical physics expert.
In order to validate these indices,
both quantities have been calculated for each image.
First of all,
ESAK has been calculated from Robson´s parametric model (5),
using technical data of the examination (kV,
mAs,
anode/filter combination) extracted from DMP and quality assurance data: tube yield and HVL for reference conditions.
Then,
AGD has been calculated from ESAK using Dance Model (3,
4).
Age of the patient,
compressed breast thickness and beam quality (HVL,
kVp) are taken into consideration in order to perform the calculation (3,
4),
following the expression:
AGD= ESAK.c.g.s [1]
Where c,
g and s are tabulated coefficients which take into account X-ray quality and glandularity.
Median values of both AGD and ESAK distributions,
calculated and indicated by the unit,
have been obtained for CC and MLO projections.
Differences for each patient and also average differences for both quantities have been analysed in order to evaluate the accuracy of the available indices.
Finally,
median AGD and median ESAK,
obtained from DMP values,
have been presented as preliminary DRLs for the facility.
Results:
Median values and standard deviation of AGD and ESAK for CC and MLO samples,
both calculated and provided by the unit,
are summarized in tables 1 and 2.
The average relative differences between them and standard deviation of differences is also presented.
The AGD value obtained with DMP is always higher than the calculated one,
with an average difference of almost 20%,
yielding a greater DRL.
Differences between calculated and indicated dose quantities have been analysed,
in order to determine the main sources of uncertainty.
First of all,
the mammographic system uses Dance model for calculation of AGD,
as stated by manufacturer,
but it does not take into account the glandularity,
and therefore it ignores coefficient c in equation [1].
Furthermore,
the real HVL and tube yield of this particular unit are not taken into account; instead,
tabulated parameters for every Inspiration unit are used. These differences entail a limitation in the use of AGD index obtained by DMP for DRL establishment.
On the other side,
ESAK values are much more alike when both methods are compared,
but in this case indicated ESAK is slighter than the calculated kerma for every patient.
The average differences are less than 6%,
so DMP programme would constitute a useful tool in order to establish ESAK based DRLs.
Hence,
a validation of each unit’s indices is mandatory prior to establishment of DRLs making use of DMP,
particularly if these values are going to take part for development of national or even local DRLs.
However,
if used as an internal optimization tool at the facility,
in order to detect some problems regarding unit operation or even staff differences,
DMP constitutes a powerful instrument.
Care must be taken,
though,
when periodically comparing doses with DRLs for optimisation purposes,
since indicated values should always be chosen.
Table 3 shows preliminary DRLs,
established for CC and MLO,
in terms of AGD and ESAK.
As it can be shown in table 3,
MLO projection DRLs are greater than for CC,
due to average larger thickness of compressed breast for that projection.