Imaging equipment and phantom
The study was conducted in a university imaging department using a Wolverson Acroma X-Ray unit (Wolverson X-ray Ltd,
Willenhall,
UK) with a Varian 130 HS X-ray tube (Varian Medical Systems,
Palo Alto,
CA) with an inherent filtration of 3 mm aluminium. An Agfa (Agfa-Gevaert,
Mortsel,
Belgium) 35 x 43 cm computed radiography (CR) image receptor was used for acquisition and images were processed using an Agfa CR 35-X digitiser.
All exposures included the use of a 10:1 reciprocating grid with (40 line/cm frequency) and a broad focal spot size of 1.2 mm. Equipment quality assurance testing,
in line with IPEM 91 (IPEM,
2005),
was performed prior to image acquisition which included an assessment of voltage accuracy,
which was found to be within tolerance.
Phantom positioning
A fixed source-to-image-distance (SID) of 115 cm was used together with automatic exposure control (AEC) using the central chamber. A tube potential of 75 kVp was selected and when combined with the above factors allowed the production of a reference image that was consistent with typical clinical imaging parameters. These exposure parameters were established following a brief consultation with four local departments and after reviewing recommendations in the EC guidelines (European Commission,
1996).
For all exposures the collimated field was adjusted to include the twelfth thoracic vertebra superiorly and the sacro-iliac joints inferiorly.
The use of fixed collimation was essential in order to ensure it did not impact on phantom radiation dose or image quality,
as the amount of scattered radiation varies when different volumes of tissue are irradiated (Fauber,
2004).
For AP projections the phantom was positioned supine in accordance with standard radiographic technique (Sloane et al.,
2010).
For PA projections the phantom was positioned prone. In order to ensure the centring point was replicated,
masking tape was applied to and wrapped around the torso of the phantom with its superior border directly at the level of the horizontal line of the AP centring point.
The diameter to the left and right of the vertical line in the AP projection was measured using a ruler and then replicated in the PA orientation.
Collimation was once again fixed and consistent with the AP projection.
Experimental technique
For each projection (AP/PA) the kVp increment was varied by 5 kVp from 75 to 110 kVp. In order to ensure continuity and minimise error the same imaging plate was used throughout the study.
Image acquisition was repeated three times for each kVp increment and at each orientation (AP/PA).
Dosimetry
Entrance surface dose (ESD) was measured using a Mult-O-Meter 407L (Unfors Instruments,
Billdal,
Sweden) positioned on the phantom at the centre of the collimated field.
In order to increase the accuracy of dose measurement the ESD was measured three times and an average value was calculated.
ESD measurements were converted to effective dose (ED) estimations using the Monte Carlo dosimetry simulation software PCXMC 2.0 (STUK,
Helsinki,
Finland).
The mean effective (ICRP 103,
2007) and absorbed doses to the stomach,
colon and remainder tissues were recorded,
as these are classified as the three most sensitive tissues irradiated during an AP lumbar spine radiography (Wall et al.,
2011).
Absorbed doses to the ovaries and testes were also recorded in order to compare findings between the two projections.
Image quality assessment
The evaluative panel consisted of five final year radiography students,
who at the time of the study were < 6 months away from qualification.
Each of the raters had previously participated in visual grading analysis (VGA) experiments and were deemed sufficiently experienced to undertake image analysis.
Images were assessed under standardised viewing conditions using two EA243WM MultiSync (NEC Corporation,
Tokyo,
Japan) 2.3 megapixel monitors. Ambient lighting,
less than 50 lx and the distance of the chair from the monitor were kept constant.
Details of how the images were acquired were blinded to all raters.
Two-alternative forced choice (2AFC) software (Hogg et al.,
2012) was used to present the acquired images to the raters. This allowed the presentation of the reference image concurrently alongside the comparator images on the monitor but in a randomised order. A further advantage of this software was that it prohibited zooming and window width or level adjustments. Previous research has reported on the benefits of 2AFC in that it permits easier detection of differences in quality when compared to an absolute method where observers are asked to evaluate images utilising criteria without a comparison reference image.
Finally,
magnification was assessed and compared between the AP and PA projections using the software program Image J (National Institute of Health,
Bethesda,
MD).
This was assessed in the same manner as that employed by Heriard,
Terry & Arnold (1993) who determined the magnification differences between the two projections by measuring the transverse diameter of the vertebral body of L3 (Tsuno & Shu,
1990).