Fig. 17.1
PMMA phantom used to perform the AEC testing
SdNR and dose should be measured in at least three different thicknesses (20, 45 and 70 mm) which are considered to mimic attenuation and scatter provided by a thin, average and large breast. The PMMA breast phantom is composed of various 0.5 or 1 cm slabs piled up on top of each other to produce the necessary thickness. A small aluminium square (1 cm × 1 cm and 0.2 mm thickness) must be positioned below the top slab at 6 cm from the chest wall edge. In the UK it is placed on top of the bottom slab and then built up with additional PMMA on top.
The PMMA phantom is placed on the perspex it is not positioned on the image receptor but on top of breast support platform with an overhang of 5 mm out from the chest wall edge and laterally centred in the image field. The radiation field size should be collimated to cover the complete phantom.
The compression paddle must be positioned in contact with the PMMA slabs and a consistent compression force is recommended e.g. 60 N. For AEC systems with options for positioning the AEC (X-ray sets associated with CR systems and some DR e.g. Hologic Dimensions) the midline position is selected and a region that would not be affected by the Aluminium square.
Images should be acquired using AEC and associated exposure settings typically used in clinical practice. Images are acquired for the three PMMA thicknesses. For the standard thickness (45 mm PMMA) the procedure should be repeated three times.
For thicknesses ≥40 mm, low attenuation material spacers can be positioned at the edges of the phantom to achieve the intended equivalent [breast] thickness. This is important because some mammography systems adjust the X-ray settings according to the detected breast thickness or compression force.
Only raw images with the processing algorithm turned off are used, acquired in a “raw”, “unprocessed” or DICOM “for processing” format depending on the system used.
For each image, measurements of the mean pixel value and its standard deviation are performed in ROIs (1 cm2) in aluminium and the surrounding background. Pixel values are corrected using STP data and SdNR is calculated:
(17.1)
Frequency
Every 6 months, or more frequently as required within the UK: Daily for Radiographers at 4 cm and monthly for 2 and 6/7 cm.
Expected Results
Using the SdNR method, the IAEA reference values can be used (Table 17.1).
Table 17.1
Acceptable (Accep.) and Achievable (Achiev.) reference levels for SdNR in mammography proposed by IAEA for thicknesses of 20, 45 and 70 mm [3]
Mammography system | Compressed breast thickness [mm] | |||||
|---|---|---|---|---|---|---|
20 | 45 | 70 | ||||
Accep. | Achiev. | Accep. | Achiev. | Accep. | Achiev. | |
GE 2000D – DR | 8.9 | 12.9 | 7.9 | 11.5 | 6.9 | 10.0 |
GE DS – DR | 8.9 | 12.9 | 7.9 | 11.5 | 6.9 | 10.0 |
GE Essential – DR | 12.7 | 18.4 | 11.3 | 16.5 | 9.9 | 14.4 |
Fuji Amulet – DR | 6.1 | 8.7 | 5.5 | 7.8 | 4.8 | 6.8 |
Siemens Inspiration – DR | 4.4 | 6.3 | 3.9 | 5.7 | 3.4 | 5.0 |
Detector Uniformity and Artefacts
Image receptor uniformity is essential and uniformity testing should be performed regularly. Uniformity problems in digital systems can be caused by inappropriate calibrations of the image field or due to artefacts caused by defects on the detector [9, 23]. There are also noted problems with the target, filters, grid and paddle if looking at the system rather than just detector.
Procedures and Materials
Uniformity can be assessed using flat field uniform images produced with an attenuated X-ray beam with a 2 mm Al foil attached to the tube port. Most manufacturers supply a large area block of PMMA which can sit over the breast support platform as an alternative to Al over the tube. The image receptor can be imaged using clinical exposure parameters to achieve an air kerma of approximately 100 μGy at the image detector. The images can be acquired either (a) without grid and without compression paddle and also without processing (raw images) or (b) with the grid to asess the system clinically. A large radiation field should be used (broad beam), typical for clinical use.
Pixel values should be corrected using STP data before making ROI measurements. The mean pixel value should be measured for 5 ROI (1 cm2 each), distributed as shown in Fig. 17.2: one at the centre of the image and the other 4 at the centre of each quadrant [22].
Fig. 17.2
Reference ROIs for uniformity measurements
Frequency
Following equipment service to tube or detector and more frequently as required by protocol. Within the UK: Every 6 months by technicians/physicists and monthly by Radiographers.
Expected Results
Mean SNR, calculated for all 5 ROIs should present a maximum deviation of ≤15 % [4].
The images can be assessed for artefacts. Image artefacts can have different origins, including client, practitioner and equipment-related.
The artefacts related to the client can be caused by motion or due to the anatomical characteristics (for instance the thin breast artefact (<20 mm) that is caused because it is possible that during compression, the paddle edges may be included at the corners of the image creating the artefact) [24].
The practitioner can introduce artefacts during the positioning of the breast, improper detector handling (CR systems) and inadequate screen cleaning procedures (CR systems) than can cause white dots due to dust and parts of the coating of the cassette [25].
The most common artefacts related to the equipment are those related to software processing errors and those that are caused by the specific architecture of the detector, namely geometric distortion due to incorrect stitching of sub-images and inhomogeneities towards the lateral sides of the image. Absence of detector calibration can also cause artefacts due to imperfections and differences in gain of each individual segments of the detector. The grid lines can also appear causing artefacts due to the stopping or slowing down of grid and also misplacement and vibration [23–25].
Frequency
Weekly, following an equipment service in which the image acquisition system was modified or following correction software [9, 23]. Within it is every 6 months.
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
