3.2 Patient Study
3.2.2 Image quality evaluation
The obtained X-ray images were evaluated using the ViewDEX software (Sahlgrenska University Hospital, Göteborg, Sweden). ViewDEX (Viewer for Digital Evaluation of X-ray Images) is a software that allows evaluation of digital images. It displays X-X-ray images along with the image quality criteria scale. The person who is evaluating X-rays selects the number or a letter on a scale that suits their image quality judgement. All of their choices are stored in a log file (Svalkvist et al., 2021).
The images we obtained were evaluated by three radiologists with more than five years of experience. For assessment we used a four-step scale with the numbers from 1 to 4, where number 4 means perfect/optimal image, 3 means good image, 2 moderate image (acceptable for diagnostic purposes) and 1 inadequate image. The radiologists were blinded to whether each image was acquired with or without fat tissue removal.
The images were evaluated according to the following image quality criteria for pelvic
visualisation of medulla and cortex of the pelvis,
visualisation of sacrum and its foramina,
visualisation of pelvic/hip soft tissues.
After image evaluation scores were received from the radiologists, the average score was calculated for each criterion for each image, so that every image had one average score for each of the criteria. The total score of each image was calculated by adding all image criteria average scores for every evaluated image.
22
3.2.3 Statistical analysis
Data were analysed using SPSS version 25 (IBM Inc, Armonk, New York). We proposed to use the Shapiro-Wilk test to determine the normal distribution of all data. For normally distributed data on patient measurements, the T-test for independent samples was used and paired samples T-test was used for dependent data, while the Mann-Whitney U-test was used for independent non-normally distributed data and Wilcoxon test was used for dependent non-normally distributed data. To estimate reliability of the radiologists we used Cohen’s Kappa coefficient. Results are presented with tables and graphs. The significance of p<0.05 was used for all the tests.
4 RESULTS
In this chapter the results of the research are presented. In the first part, the results of the phantom study will be reported in which the appropriate type of displacement band that is best suited for removing fat tissue from the area of interest during erect pelvic radiography.
It was important that the band did not cause artefacts on the images. In the second part of this chapter, the results of the patient study which compared the image quality ratings and radiation doses received between patients who had the fat tissue removed from the field of view during erect pelvic radiography and those who did not were reported.
4.1 The results of the band comparison (phantom study)
Four bands made of different materials were evaluated. The aim was to test which one was superior in the removal of fat tissue and did not cause any artefacts on the image. A rubber exercise band, a cotton cloth folded into a band, an elastic cotton bandage, and a thin cotton triangular bandage were compared (Figure 3). Four images were acquired, one for each band.
Figure 3: A rubber exercise band (upper left corner), a cotton cloth folded into a band (upper right corner), an elastic cotton bandage (lower left corner) and a thin cotton
triangular bandage (lower right corner)
24
Members of the research team (student and supervisors) were responsible for visually evaluating the level of soft tissue displacement and the presence of any related image artefacts. Each band was deemed by consensus adequate to move the simulated fat tissue from the pelvic area, but only one did not show any artefacts on the radiographic image. The cotton cloth folded into the band was most visible on the image. An exercise rubber band and an elastic cotton bandage also caused visible artefacts on the image, but not as much as a cotton cloth. Finally, a thin cotton triangular bandage did not show any visible artefacts, so this was used for the second part of the study.
4.2 The results of the patient study
This part of the study was carried out on 60 patients, of whom 34 were women and 26 men.
The average ± SD age was 67 ± 11 years. The mean ± SD age of the females was 68 ± 12 years and that of the males 67 ± 11 years.
Patients were randomly divided into two equal groups. One half of the participants were in a group where they had to move the fatty tissue from the ROI during the pelvic imaging. For this, they used a thin cotton bandage, which was deemed in the phantom study as the most suitable for lifting the soft (fat) tissue and least visible. Meanwhile, the patients in the second group did not move the tissue away during erect pelvic radiography.
4.2.1 Calculated body mass index of the patients
Out of 60 patients, 22 (36.7%) were overweight according to the BMI classification, 34 (56.7%) had class I obesity and 4 (6.7%) of them had class II obesity. In the group where participants did not have to move fat tissue, the average ± SD BMI was 30.8 ± 3.0 and in the other group where the tissue was moved, the average ± SD BMI was 31.0 ± 2.7. The statistical analysis of body mass index between two inspected groups is shown in Table 5.
Table 5: Statistical analysis of BMI for group of patients with and without fat tissue removal
Group n Mean ± SD. Median Min. Max. p-value Without tissue removal 30 30.8 ± 3.0 31.2 25.5 36.6 0.790
With tissue removal 30 31.0 ± 2.7 30.7 27.2 38.8
The difference between the compared groups according to BMI values is minimal (0.6%).
Using the T-test for independent samples, it was found that there was no statistically significant difference in BMI between the patients who removed fat tissue and those who did not (p=0.790). The graphical representation of the distributed data within the groups is shown in Figure 4.
Figure 4: Distributed BMI values in groups divided by tissue removal
4.2.2 Measured waist and hip circumference of the patients
In the group of patients with the fat tissue removal, measurements of waist and hip circumference were taken twice, i.e., the first time before and the second time after tissue removal. In the other group, measurements were taken only once. The average ± SD waist circumference in that group was 112.9 ± 7.4 cm and hip circumference was 108.7 ± 7.2 cm (Table 6). Patients who had to remove fat tissue during imaging had mean ± SD waist and hip circumferences of 111.7 ± 5.3 cm and 105.6 ± 6.1 cm, respectively. Raising fat tissue from the region of interest decreased waist circumference to an average of 106.4 ± 5.2 cm, but hip circumference did not change (105.5 ± 6.0 cm). Waist thickness decreased by 4.7%
and hip thickness by 0.1%. The main statistical analysis is shown in Table 7.
26
Table 6: Statistical analysis of waist and hip circumference between both groups of patients
Using the independent samples T-test there were no statistically significant differences between waist (p=0.459) or hip circumference (p=0.077) of the patients who removed fat tissue during imaging and those who did not.
Table 7: Statistical analysis of waist and hip circumference before and after fat tissue removal
Using paired samples T-test, it was found that waist circumference before and after tissue removal was statistically different (p<0.001), as the thickness decreased by 4.7%, while hip circumference before and after removal was not statistically different (p=0.211).
4.2.3 Primary field size measurements in patients
The primary field size was measured in all patients. The mean ± SD field size in the group of patients without tissue removal was 1600.6 ± 144.6 cm2 and in the group with fat tissue removal was 1604.3 ± 113.4 cm2. The rest of the statistical analysis is shown in Table 8.
Table 8: Statistical analysis of primary field size for group of patients with and without fat tissue removal
Group n Mean ± SD. Median Min. Max. p-value Without tissue removal 30 1600.6 ± 144.6 1583.6 1276.8 1831.8 0.743
With tissue removal 30 1604.3 ± 113.4 1583.6 1373.9 1831.8
Using the Mann-Whitney U test, no statistically significant differences in primary field size were found between the groups of patients who removed fat tissue during pelvic imaging and those who did not (p=0.743). The distribution of primary field size data in the groups is illustrated using boxplots in Figure 5.
Figure 5: The distribution of primary field size data in groups divided by tissue removal
4.2.4 Dose area product measurements in patients
DAP was measured using the ionization chamber, which is already built into the X-ray unit.
Its value is displayed on the computer monitor after exposure for each patient. The DAP values were compared between the groups of patients, with and without fat tissue removal, and are reported in Table 9. The average ± SD DAP in the group without tissue removal was 194.4 ± 106.6 μGy m2 and was lower in the tissue removal group (119.8 ± 43.2 μGy m2), this was 38.5% lower than in the first group (p=0.001).
Table 9: Statistical analysis of DAP for group of patients with and without fat tissue removal
Group n Mean ± SD. Median Min. Max. p-value Without tissue removal 30 194.4 ± 106.6 162.3 65.6 473.3 0.001
With tissue removal 30 119.8 ± 43.2 115.8 58.0 229.2
28
The Mann-Whitney U test was used to compare the differences in DAP values between groups and concluded that there were statistically significant differences (p=0.001) in DAP between patients who removed fat tissue during imaging and those who had not. The comparison of the distribution of DAP data between the groups is shown in Figure 6.
Figure 6: Distributed DAP values in groups divided by tissue removal
4.2.5 Entrance skin dose
The average ± SD ESD for patients in a group without fat tissue removal was 5.12 ± 2.68 mGy and in a group with fat tissue removal was 2.87 ± 1.05 mGy (p<0.001).
The ESD was 44% lower in the group in which patients moved the fat tissue from the region of interest. Statistical analysis of ESD for each group is shown in Table 10.
Table 10: Statistical analysis of ESD for group of patients with and without fat tissue removal
Group n Mean ± SD. Median Min. Max. p-value Without tissue removal 30 5.12 ± 2.68 4.57 1.62 11.77 p<0.001
With tissue removal 30 2.87 ± 1.05 2.78 1.39 5.56
Using the Mann-Whitney U test, the differences between the ESD calculations for each group were compared. Statistically significant differences between the groups in the calculated entrance skin doses were identified (p<0.001). The comparison of the ESD values between the groups is shown in Figure 7.
Figure 7: Distributed ESD values in groups divided by tissue removal
4.2.6 Effective dose
Effective doses between groups are compared in Table 11. The average ± SD effective dose in a group of patients without tissue removal was 237.8 ± 113.2 μSv and in a group of patients with tissue removal was 145.8 ± 46.1 μSv (p<0.001). The effective dose was reduced by 38.7% in a group of patients with soft tissue removal.
Table 11: Statistical analysis of effective dose for group of patients with and without fat tissue removal
Group n Mean ± SD. Median Min. Max. p-value Without tissue removal 30 237.8 ± 113.2 207.5 98.4 528.3 p<0.001
With tissue removal 30 145.8 ± 46.1 144.7 79.8 270.6
To compare the differences of effective dose between two groups we used Mann-Whitney U test. We found statistically significant differences between group of patients who removed soft tissue from the region of interest and those who did not (p<0.001). The comparison of distribution of effective dose values in each group are presented in Figure 8.
30
Figure 8: Distributed effective dose values in groups divided by tissue removal
4.2.7 Results of the image quality evaluations
Images obtained during the patient study were evaluated by three radiologists. Each radiologist evaluated all 60 images using the ViewDEX software. They evaluated the images using image quality criteria with a four-point scale. For each image, they evaluated the visualisation of the hip joints, trochanters, sacroiliac joints, iliac crests, acetabula, pubic/ischial rami, femoral necks, medulla and cortex of the pelvis, sacrum, its foramina, and surrounding soft tissues. Each criterion was scored with a number from 1 to 4, with the number 4 being the highest score, meaning that each image could receive a maximum score of 40. All images contained all of the required anatomical points.
4.2.7.1 The reliability of radiologists
We calculated the reliability of radiologists using Cohen’s Kappa coefficient. We estimated the level of agreement between radiologists in given scores for each criterion, the higher the Kappa coefficient is, stronger the agreement is. The reliability between radiologist number 1 and 2 was generally low, as the coefficient for each criterion did not exceed value 0.40 (good level of agreement). The highest level of agreement between radiologist 1 and 2 were in determining the visualisation of trochanters (κ=0.274), which means there was fair agreement between them. They least agreed in scores for the visualisation of medulla and
cortex of the pelvis, where their level of agreement was estimated with -0.001, which means there was no agreement. The rest of the Kappa coefficient values of the reliability of radiologist number 1 and 2 are presented in Table 12.
Table 12: The reliability between radiologist 1 and 2
Criterion Kappa coefficient
Visualisation of medulla and cortex of the pelvis. -0.001 Visualisation of sacrum and its foramina. 0.025
Visualisation of pelvic/hip soft tissues. 0.005
The reliability between radiologist number 1 and 3 was also low in general. They had the higher agreement in scores for the criterion of visualisation of trochanters (κ=0.272) and iliac crests (κ=0.258), but these values show the fair agreement between them. They had no agreement when evaluating the visualisation of femoral necks, sacrum and its foramina and pelvic/hip soft tissues, as the Kappa values were -0.032, -0.006 and -0.050, respectively. The reliability of radiologist number 1 and 3 is presented in Table 13.
Table 13: The reliability between radiologist 1 and 3
Criterion Kappa coefficient
Visualisation of medulla and cortex of the pelvis. 0.000 Visualisation of sacrum and its foramina. -0.006
Visualisation of pelvic/hip soft tissues. -0.050
32
Out of all three radiologists, radiologist number 2 and 3 had the higher level of agreement.
They most agreed in evaluating the visualisation of hip joints (κ=0.321), which means there was fair level of agreement. There was no agreement in evaluation of visualisation of pelvic/hip soft tissues (κ=-0.042). The rest of the Kappa values for each criterion is presented in Table 14.
Table 14: The reliability between radiologist 2 and 3
Criterion Kappa coefficient
Visualisation of medulla and cortex of the pelvis. 0.099 Visualisation of sacrum and its foramina. 0.086 Visualisation of pelvic/hip soft tissues. -0.042
4.2.7.2 Visualisation of hip joints
The median of scores in the assessment of visualisation of the hip joints was 3.7 (IQR=1.0) in the group without fat tissue removal and 4.0 (IQR=0.1) in the group with fat tissue removal (p=0.001). The minimum score in the first group was 2.7 and in the second group was 3.0.
In both groups, the maximum score was 4 (Table 15).
Table 15: Statistical analysis of evaluation score of visualisation of hip joints in groups with and without fat tissue removal
Group n Mean ± SD. Median Min. Max. p-value Without tissue removal 30 3.5 ± 0.5 3.7 2.7 4.0 0.001
With tissue removal 30 3.9 ± 0.3 4.0 3.0 4.0
Using the Mann-Whitney U test, we found statistically significant differences in hip visualisation assessment scores between groups (p=0.001). The comparison between the groups is shown in Figure 9.
Figure 9: The distribution of evaluation scores for visualisation of hip joints in groups divided by tissue removal
4.2.7.3 Visualisation of trochanters
The trochanter visualisation assessment had a median at 3.2 (IQR=1.5) in the group in which the patients did not move fat tissue during the examination and 3.7 (IQR=1.0) in the group of patients who did (p=0.021). The minimum and maximum scores in the first group were 1 and 4, respectively, and 1.3 and 4 in the second group. The rest of the statistical analysis is shown in Table 16.
Table 16: Statistical analysis of evaluation score of visualisation of trochanters in groups with and without fat tissue removal
Group n Mean ± SD. Median Min. Max. p-value Without tissue removal 30 2.9 ± 1.0 3.2 1.0 4.0 0.021
With tissue removal 30 3.4 ± 0.8 3.7 1.3 4.0
The Mann-Whitney U test was used to compare differences between trochanter visualisation assessment scores for each group. There were statistically significant differences in the assessment scores between the groups with and without adipose tissue removal (p=0.021) (Figure 10).
34
Figure 10: The distribution of evaluation scores for visualisation of trochanters in groups divided by tissue removal
4.2.7.4 Visualisation of sacroiliac joints
When visualisation of the sacroiliac joints was evaluated, the median in the group without fat tissue removal was at 3.2 (IQR=1.1) and in the group with tissue removal was at 3.7 with IQR at 1.0 (p=0.055). The minimum score in the first group of patients was 1.3 and in the second group was 2.7. The maximum score in both groups was 4. The statistical analysis is shown in Table 17.
Table 17: Statistical analysis of evaluation score of visualisation of sacroiliac joints in groups with and without fat tissue removal
Group n Mean ± SD. Median Min. Max. p-value Without tissue removal 30 3.1 ± 0.8 3.2 1.3 4.0 0.055
With tissue removal 30 3.5 ± 0.5 3.7 2.7 4.0
Based on the result of the Mann-Whitney U test the p-value was slightly above the borderline of statistical significance (p=0.055). We concluded there were no statistically significant differences in scores between groups in the assessment of sacroiliac joint visualisation. The comparison between the groups is shown in Figure 11.
Figure 11: The distribution of evaluation scores for visualisation of sacroiliac joints in groups divided by tissue removal
4.2.7.5 Visualisation of iliac crests
When the visualisation of the iliac crests was evaluated, the median was at 3.7 (IQR=0.8) in the group of patients without tissue removal and 4.0 (IQR=0.4) in the group of patients with fat tissue removal (p=0.060). The minimum score was 2.3 in the first group and 3.0 in the second group. In both groups, the maximum score was 4. The rest of the statistical analysis can be found in Table 18.
Table 18: Statistical analysis of evaluation score of visualisation of iliac crests in groups with and without fat tissue removal
Group n Mean ± SD. Median Min. Max. p-value Without tissue removal 30 3.5 ± 0.5 3.7 2.3 4.0 0.06
With tissue removal 30 3.7 ± 0.3 4.0 3.0 4.0
Using the Mann-Whitney U test, the pelvic crest visualisation assessment score was also at the borderline level of statistical significance (p=0.060), but we considered this result like there were no statistically significant differences between groups. The comparison of the distribution of the score values between the groups is shown in Figure 12.
36
Figure 12: The distribution of evaluation scores for visualisation of iliac crests in groups divided by tissue removal
4.2.7.6 Visualisation of acetabula
Visualisation of the acetabula was assessed with median at 3.7 (IQR=0.7) in the group without fat tissue removal and 4.0 (IQR=0.3) with fat tissue removal (p<0.001). The minimum score was 2.3 in the first group and 3 in the second. The maximum score was 4 in both groups. The statistical analysis is shown in Table 19.
Table 19: Statistical analysis of evaluation score of visualisation of acetabula in groups with and without fat tissue removal
Group n Mean ± SD. Median Min. Max. p-value Without tissue removal 30 3.5 ± 0.5 3.7 2.3 4.0 p<0.001
With tissue removal 30 3.9 ± 0.2 4.0 3.0 4.0
Mann-Whitney U-test showed that there were statistically significant differences in the evaluation scores for acetabula visualisation when comparing the two groups (p<0.001). The distribution of assessment scores in the groups are illustrated in boxplots as seen in Figure 13.
Figure 13: The distribution of evaluation scores for visualisation of acetabula in groups divided by tissue removal
4.2.7.7 Visualisation of pubic/ischial rami
The radiologists rated the visualisation of the pubic/ischial rami in the group without tissue removal with median at 3.7 (IQR=0.7) and in the group in which the patients had the fat tissue displaced with a score of 4.0 and IQR at 0.3 (p=0.166). The minimum score was 2.7 and 3.0 for the first and second groups, respectively, and the maximum score for both groups was 4 (Table 20).
Table 20: Statistical analysis of evaluation score of visualisation of pubic/ischial rami in groups with and without fat tissue removal
Group n Mean ± SD Median Min. Max. p-value Without tissue removal 30 3.7 ± 0.3 3.7 2.7 4.0 0.166
With tissue removal 30 3.8 ± 0.3 4.0 3.0 4.0
There were no statistically significant differences between the groups and the evaluation results of the visualisation of the pubic/ischial rami, based on the results of the
There were no statistically significant differences between the groups and the evaluation results of the visualisation of the pubic/ischial rami, based on the results of the