|
Introduction
Dynamic cueing is an effective way of stimulating perception of regions of interest within radiological images. This
study explores the impact of a novel teaching tool using dynamic cueing for lesion detection on plain chest radiographs.
Materials and methods
Observer performance studies were carried out where 36 novices examined 30 chest images in random order. Half of
these contained between one and three simulated pulmonary nodules. Three groups were investigated: A (control: no
teaching tool), B (retested immediately after undergoing the teaching tool) and C (retested a week after undergoing the
teaching tool). The teaching tool involved dynamically displaying the same images with and without lesions. Results
were compared using Receiver Operating Characteristics (ROC), sensitivity and specificity analyses.
Results
The second reading showed significantly greater area under the ROC curve (Az value) (p<0.0001) and higher sensitivity
value (p=0.004) compared to the first reading for Group B. No differences between readings were demonstrated for
groups A or C. When the magnitudes of the above changes were compared between Group B and the other two groups,
greater changes in Az value for Group B were noted (B vs. A:p=0.0003, B vs. C:p=0.0005). For sensitivity, when Group
B was compared to Group A, the magnitude of the change was significantly greater (p=0.0029) whereas when Group B
was compared to Group C, the magnitude change demonstrated a level approaching significance (p=0.0768).
Conclusions
The novel teaching tool improves identification of pulmonary nodular lesions on chest radiographs in the short term.
|
