Using signal detection theory to model changes in serial learning of radiological image interpretation

Area of Science:

• Medical Education
• Radiology
• Cognitive Psychology

Background:

• Signal detection theory (SDT) provides metrics (d' for discrimination, λ for criterion) to assess performance in tasks involving distinguishing signals from noise.
• Understanding how these parameters change with experience is crucial for optimizing learning in diagnostic fields like radiology.

Purpose of the Study:

• To investigate the serial changes in SDT parameters (d' and λ) as learners are repeatedly exposed to radiological cases.
• To compare the learning trajectories of individuals with varying levels of experience in interpreting radiographs.

Main Methods:

• 46 participants (medical students, residents, fellows, staff emergency physicians, and radiologists) were exposed to 234 ankle radiographs via a web-based application.
• Participants classified cases as normal or abnormal, identified abnormality locations, and received immediate feedback.
• Serial changes in discrimination ability (d') and response criterion (λ) were analyzed across experience groups.

Main Results:

• Lower experience groups (students, residents, fellows) demonstrated progressive improvement in discrimination ability (d') with case exposure.
• Higher experience groups (emergency physicians, radiologists) exhibited consistently high and stable discrimination ability.
• Differences in response criterion (λ) were observed, with less experienced groups initially showing a tendency towards undercalling abnormalities (more false negatives), which diminished with increasing expertise.

Conclusions:

• SDT metrics effectively capture learning-related changes in radiograph interpretation skills.
• The findings suggest that SDT parameters can inform the development of tailored instructional strategies for different learner groups in radiology education.