CHEST Guidelines-Response

Response_chest (29)

Pdf Summary

In response to insightful comments by Dr. McGaghie regarding published articles on chest radiography (CXR), the authors address the issue of perceptual errors as a significant cause of missed findings in CXR interpretations. They concur that perceptual training methods, while still under investigation, represent a promising approach to mitigating such errors. Traditional apprentice-style training in CXR is considered slow and incomplete, suggesting that it can be enhanced through deliberate practice. This method involves focused, repetitive, rapid, and targeted training with immediate feedback to refine skills.<br /><br />Perceptual training in radiology, introduced decades ago by Kundel and Nodine, utilizes computer-based learning and simulations that mimic clinical environments. Trainees practice identifying radiographic abnormalities in quick succession with instant feedback, and this game-like training is comparable to the reinforcement learning processes in machine learning.<br /><br />The benefits of perceptual training are highlighted by a study on detecting femoral neck fractures. It shows that even non-specialist individuals, after under an hour of deliberate perceptual training, can perform comparably to board-certified radiologists and artificial intelligence models. Acknowledging the rapid learning capability of humans, the authors suggest that AI training data could simultaneously be deployed to train radiologists, enhancing their lesion detection skills.<br /><br />In conclusion, the authors advocate for including deliberate practice in the arsenal of strategies to reduce error in CXR interpretation, supporting Dr. McGaghie's recommendations. The letter, authored by Warren B. Gefter and Hiroto Hatabu from leading radiology departments, emphasizes the importance of integrating perceptual training into broader educational frameworks to bolster radiologist accuracy and proficiency.

Keywords

perceptual errors

CXR interpretation

deliberate practice

perceptual training

radiology education

computer-based learning

instant feedback

AI training data

radiographic abnormalities

lesion detection