Radiology's Achilles' heel: error and variation in the interpretation of the Röntgen image.

Publisher Website
Google Scholar
Abstract
The performance of the human eye and brain has failed to keep pace with the enormous technical progress in the first full century of radiology. Errors and variations in interpretation now represent the weakest aspect of clinical imaging. Those interpretations which differ from the consensus view of a panel of "experts" may be regarded as errors; where experts fail to achieve consensus, differing reports are regarded as "observer variation". Errors arise from poor technique, failures of perception, lack of knowledge and misjudgments. Observer variation is substantial and should be taken into account when different diagnostic methods are compared; in many cases the difference between observers outweighs the difference between techniques. Strategies for reducing error include attention to viewing conditions, training of the observers, availability of previous films and relevant clinical data, dual or multiple reporting, standardization of terminology and report format, and assistance from computers. Digital acquisition and display will probably not affect observer variation but the performance of radiologists, as measured by receiver operating characteristic (ROC) analysis, may be improved by computer-directed search for specific image features. Other current developments show that where image features can be comprehensively described, computer analysis can replace the perception function of the observer, whilst the function of interpretation can in some cases be performed better by artificial neural networks. However, computer-assisted diagnosis is still in its infancy and complete replacement of the human observer is as yet a remote possibility.