Keywords:
Education and training, Diagnostic procedure, Digital radiography, Musculoskeletal bone
Authors:
K. McNally, K. Matthews; Dublin/IE
DOI:
10.26044/ecr2019/C-3387
Methods and materials
Focus groups
Focus groups were used initially to discover what variations in centring points are taught to students in clinical practice.
Twelve final year undergraduate Radiography students volunteered to participate and were divided randomly into four groups.
Using a set of trigger questions,
the groups identified four projections with most noticeable variations in taught centring points: lateral lumbar spine,
lateral knee,
antero posterior (AP) shoulder and oblique lateral hip projections.
Details of the centring points reported by the focus groups were recorded,
along with the published centring point of each projection from authoritative texts (Table 1, Table 2, Table 3, Table 4).
X-ray study
Following ethical review and approval of the study,
the centring points were applied to a cadaver in the x-ray room of the university.
One image was produced for each centring point,
resulting in a total of 20 images.
Each image was given a unique code to permit appraisal independent of knowledge of the centring point.
Image evaluation
The 20 images were evaluated using visual grading analysis (VGA).
For each projection,
the images produced with the published centring point were used as the reference image.
The image elements most likely to be affected by centring were considered to be distortion of anatomy,
clearance of anatomy from overlying structures,
accuracy of collimation and automatic exposure control (AEC) operation [3,
6-8].
Anatomical criteria to permit evaluation of these elements were developed for each projection: an example for the lateral knee projection is shown in Table 5.
Data analysis
IBM SPSS was used for statistical analysis.
To evaluate the suitability of the selected grading criteria,
a sample of eight images was scored by two independent observers.
Spearman’s test of correlation between observers was low (rho = 0.225,
p=0.058), revealing some criteria lacked objectivity.
These were modified and re-tested and high correlation between observers was achieved (rho = 0.832,
p=0.000).
Once the criteria were finalised,
each image was evaluated in comparison with its reference image and given an image quality score.