Remote and Automated QC
Remote and automated QC consists of different components and responsibilities,
as described in Figure 1.
Remote QC focuses on analyzing an entire image rather than using localized simple measurements made manually on the image.
It consists of the following major components: local image acquisition,
local image verification and artefact analysis,
image upload,
centralized image analysis and result analysis,
reporting and feedback.
Automated QC consists of the following major components: local data acquisition,
local image verification and artefact analysis,
local automated image analysis,
data upload,
centralized results analysis,
reporting and feedback.
The measurements required for the automated QC are meant to be performed either automatically or by the local personnel.
For facilities without image transmission capabilities:
The simplest model for implementing a remote QC program is through the transmission of numerical data obtained from measurements,
without the need to transmit the associated images.
In this case,
the analysis of the images and the collection of relevant data takes place locally (e.g.
with the ATIA application).
The results obtained from those measurements are transmitted to the supervising CQMP in order to analyze the system performance over time at the pre-established testing frequency.
While simplest,
this method is very limited in the QC measurements that can be performed.
For facilities with image transmission capabilities:
Actual transmission of the quality control images is certainly a much more complicated task,
but it is also a means to provide a more comprehensive and sustainable solution when compared to the transmission of data only.
Description of phantoms
Phantoms for radiography and mammography that are proposed in the IAEA document can be easily built locally.
These simple phantoms generate a large,
uniform field for artefact analysis by means of a uniform attenuator.
Each phantom has also simple square targets in a test plate that allow for densitometric evaluation of analogue films and for advanced analysis of digital images.
Radiographic phantom:
The radiographic phantom consists of two parts:
1) A target plate made of a 28 cm × 28 cm × 0.5 cm of PMMA.
This base plate may be multiple thinner sheets of PMMA stacked together to achieve a total thickness of 0.5 cm,
if this is more cost effective.
The targets are a 5 cm × 5 cm × 0.2 cm Cu square and a 1 cm × 1 cm × 0.4 cm Al square which lay flat on the target plate.
The positions of both test objects can be seen in Figure 2 and 3.
The copper square must be angled approximately 3° with respect to the edge of the target plate to enable measurement of MTF.
2) A 10 cm × cm 10 × 0.2 cm copper plate that serves as the main attenuator for the phantom.
Several thinner sheets of copper may be combined for a total 0.2 cm,
if this is more cost effective.
Mammography phantom:
Similar to the radiographic phantom,
the mammography phantom also consists of two parts:
(1) A target plate made of a 24 cm × 30 cm × 0.5 cm sheet of PMMA.
The targets are a 5 cm × 5 cm × 0.1 cm Cu square and a 1cm × 1 cm × 0.02 cm Al square which lay flat on the target plate.
The positions of both test objects can be seen in Figure 4 and 5.
The copper square is angled approximately 3° with respect to the edges of the target plate for an accurate determination of the MTF.
(2) A 24 cm × 30 cm × 4 cm PMMA plate that serves as the main attenuator for the phantom.
This base plate may consist of multiple thinner plates stacked together,
if this is more cost effective.
The copper squares are used for densitometric measurements in analogue systems and for Modulation Transfer Function (MTF) and detectability index [7,8] analysis in digital systems.
The detectability index,
d’ metric has been developed,
which relates subjective measurements such as SNR and MTF to actual,
clinical interpretation tasks [9,10,11].
Through d’,
a simple phantom can give direct relationship to clinical imaging performance.
Angling the copper squares relative to the target plate also angles it relative to the digital matrix,
thereby eliminating aliasing artifacts in the analysis.
The aluminium squares are used for densitometric measurements in analogue systems and for Contrast to Noise Ratio (CNR),
Signal Difference to Noise Ratio (SDNR),
and d-prime (dꞌ) analysis in digital systems.
Figures 6 and 7,
show images produced with the proposed phantoms.
Automated Tool for Image Analysis (ATIA)
The Automated Tool for Image Analysis (ATIA) was developed in order to facilitate the analysis of radiography and mammography phantom images described in the IAEA publication under development.
ATIA has been specifically designed for the needs of automated and remote QC as described in the context of these guidelines.
To avoid the subjective nature of traditional metrics,
advanced metrics such as SDNR,
MTF and detectability index (dꞌ) are calculated by the ATIA application.
None of these metrics is dependent upon the observer,
so the impact of different individuals imaging the test tool is negligible.
ATIA software provides the user with an CSV data file (Figure 8) with all relevant results that can be opened with any spreadsheet software or another application; with this file,
trend analysis can be performed by the CQMP.
The ATIA application includes a function for highlighting areas of non-uniformity and artefacts (Figure 9 and 10).
This function may be run on the image with the test plate,
recognizing that the test targets will be identified by the application,
or it can be run on a separate,
uniform image with only the base attenuator.
In either case,
images should be visually reviewed by either the facility or the CQMP to ensure that artefacts and non-uniformities highlighted by the application are not overlooked.
Image analysis of the phantom images can also be done with freely downloadable software such as ImageJ,
which is extensively used by the scientific community for the measurement and analysis of medical images.