59
60
Based on the data obtained, it was found that the major difficulty was caused by Case N4 (Figure 22) with pathology localization in the form of ground-glass pulmonary compaction in the apical section of the left lung according to CT and visualized behind the shadow of the collarbone on a PA chest X-ray (Table 8). This case was incorrectly interpreted by 61.3% of radiologists, regardless of work experience and exposure to thoracic radiology.
Figure 22: Case N4, patient with adenocarcinoma in C1+2 of the left lung
Also, when localizing pathological changes behind the shadow of the ribs on PA chest X-rays, as in Cases N2 and N3, 33.3% and 37.3% of wrong X-ray interpretations were obtained respectively. At the same time, errors were revealed in 34.4% of the X- ray interpretations among doctors with less than 10 years of work experience.
The highest percentage of correct answers was obtained in the analysis of digital X-ray images with pathology localization in the form of solid nodules and masses in the upper and lower pulmonary lobes as shown on CTs, i.e., Cases N1 and N5 (amounted to 93.3% and 92% respectively) among the X-ray analysis results by radiologists, without significant differences depending on the work experience and exposure to thoracic radiology.
The next step in researching the interpretation quality of digital X-rays of radiologists was testing the radiologists with an online platform; the testing gathered 516 specialists [137].
Рекомендовано к изучению разделом по лучевой диагностике сайта https://meduniver.com/
61
The specialists were divided into 5 groups based on their work experience: up to 1 year (N=56), 1-2 years of work experience (N=179), 3-5 years of work experience (N=143), 6-10 years (N=87) and more than 10 years of work experience (N=51) [137].
62
Рекомендовано к изучению разделом по лучевой диагностике сайта https://meduniver.com/
63
64
The highest sensitivity in pulmonary nodules and masses detecting (75%) is determined among the aspiring doctors (with work experience of less than a year) with a constant decrease as work experience increases, reaching a minimum (50%) among the doctors with more than 10 years of experience (Table 9) [137]. Thus, the highest rate of omitted pathology was recorded among the radiologists with more than 10 years of work experience (44.12%), decreasing when the work years decrease and reaching 31.5% among radiologists at the very beginning of their professional career [137]. Similar results were obtained when researching cases with correct interpretation of digital X-rays with no pathology; radiologists rarely correctly see images as norm when they have work experience of 1–2 years, in 78.3% of cases. At the same time, this indicator gradually increases with when work experience increases, reaching 84.1% for doctors who have been working for more than 10 years.
Based on the online testing results, as well as with the in-person testing, no significant difference in terms of diagnostic efficiency depending on the work experience of a radiologist was obtained.
Figure 23: Area under curve for radiologists (mean value)
The characterological curve results from the online testing were similar to the characterological curve results from in-person testing (AUC of 0.759, AUC of 0.722), falling short of the required threshold (Figure 23).
Рекомендовано к изучению разделом по лучевой диагностике сайта https://meduniver.com/
65
After analyzing the research results, it was revealed that the highest rate of omitted pathology was recorded among the radiologists with more than 10 years of work experience (44.12%), decreasing when the work years decrease and reaching 31.5% among radiologists at the very beginning of their professional career. Similar results were obtained when researching cases with correct interpretation of digital X- rays with no pathology; radiologists rarely correctly see images as norm when they have work experience of 1–2 years, in 78.3% of cases. At the same time, this indicator gradually increases with when work experience increases, reaching 84.1% for doctors who have been working for more than 10 years.
The results obtained also affect the gradual decrease in the sensitivity value and the increase in the specificity index when work experience of a radiologist increases on the contrary. Thus, the lowest sensitivity values were among the radiologists with more than 10 years of experience (50%), while the specificity value in this group was 88.3% [137]. The highest sensitivity value was observed among the radiologists with work experience up to five years (66.7%), while the specificity value amounted to 88.3%. The positive predictive value turned out to be higher among the radiologists with three to five years of experience (21.6%), as well as the likelihood ratio of a positive test (3.5). The negative predictive value was slightly higher among the radiologists with less than five years of experience, averaging at 97.4%. The likelihood ratio of a negative result is higher for the radiologists with more than 10 years of experience (0.6). The rates of under-diagnosis averaged at 37.6%, with the lowest value among the doctors with less than three years of experience (33.3%) and the highest value among the doctors with more than ten years of experience (44.1%) [137].
66
Figure 24: Area under curve for radiologists with experience of up to 10 years (a), with experience of more than 10 years (b)
The diagnostic efficiency values comply with the results obtained during the inperson testing, decreasing when the work experience of a radiologists increases (Figure 24).
As with the in-person testing, we analyzed the interpretation results of digital chest X-rays of radiologists, depending on their experience in thoracic radiology. Also, the specialists were divided into two categories: radiologists exposed to thoracic radiology (N=130) and radiologists with no such exposure (N=386) [137].
Table 10: Efficiency in detecting pulmonary nodules and masses depending on work experience in thoracic radiology
|
Results of doctors with |
Results of doctors |
|
|
|
without exposure to |
|
||
Parameters |
exposure to thoracic |
p |
||
thoracic radiology |
||||
|
radiology (N=130) |
|
||
|
(N=386) |
|
||
|
|
|
||
Sensitivity (%) |
66.7 |
66.6 |
0.012 |
|
[33.3; 83.3] |
[50.0; 87.5] |
|||
|
|
|||
Specificity (%) |
83.5 |
86.2 |
0.017 |
|
[69.1; 93.7] |
[72.3; 97.9] |
|||
|
|
|||
Likelihood Ratio of a |
3.0 |
3.0 |
0.793 |
|
Positive Test (un.) |
[2.0; 5.9] |
[1.9; 5.2] |
||
|
||||
Likelihood Ratio of a |
0.5 |
0.5 |
0.038 |
|
Negative Test (un.) |
[0.3; 0.7] |
[0.2; 0.7] |
||
|
||||
Positive Predictive Value |
16.7 |
18.5 |
0.007 |
|
(%) |
[11.6; 28.6] |
[12.1; 50.0] |
||
|
||||
Negative Predictive Value |
96.8 |
97.2 |
0.028 |
Рекомендовано к изучению разделом по лучевой диагностике сайта https://meduniver.com/
67
(%) |
[95.7; 98.3] |
[95.8; 98.8] |
|
|
Accuracy (%) |
82.0 |
85.0 |
0.014 |
|
[69.8; 91.0] |
[72.0; 94.0] |
|||
|
|
As opposed to the testing results from the previous sampling package (75 specialists), radiologists with experience in thoracic radiology did not show better results when analyzing digital X-rays as compared to their colleagues without such experience and even conceded to them in terms of some of the diagnostic efficiency values (Figure 25). Thus, the sensitivity value for doctors exposed to thoracic radiology was 57.8%, while for the doctors without such exposure it amounted to 64.9% (Table 10). The specificity value was also higher for doctors without exposure to thoracic radiology and amounted to 82.1%. According to the results, the specialists exposed to thoracic radiology missed 42.2% of cases with existing pathology on digital X-rays and falsely interpreted 20.9% cases normal X-rays as those with pulmonary pathology. Their colleagues without the said work experience received the under-diagnosis value of 35% and over-diagnosis value of 17.9%.
Figure 25: Area under curve for radiologists with experience in thoracic radiology (a), without said experience (b)
Table 11: Summary data on correct answers in the in-person and online testing of radiologists
Correct Answers, |
Total |
Up to 10 |
More than |
No work |
With work |
|
% |
years |
10 years |
experience |
experience |
||
|
||||||
In-person testing |
73.3 |
73.1 |
72.5 |
72.5 |
75 |
|
Online testing |
80.2 |
80.0 |
82.4 |
81.1 |
77.8 |
68
Рекомендовано к изучению разделом по лучевой диагностике сайта https://meduniver.com/