Abstract
OBJECTIVE: To facilitate storage, retrieval, and analysis of radiographic images we assessed the validity of a film digitizer and computer based system. METHODS: A total of 101 sets of radiographs of hands and feet from patients with early rheumatoid arthritis were digitized according to standards defined by the American College of Radiology. Two blinded observers applied the modified Larsen score to the digitized images and the original radiographs. RESULTS: There was good concordance between the scoring of digitized images and radiographic films; intraclass correlation coefficient was 0.89. Regression analysis of a Bland-Altman plot revealed that there was no significant systematic bias. It was noted that scoring the digital images took more than twice as long as scoring the radiographic films. CONCLUSION: Our results indicate that modified Larsen scoring of adequately digitized films on standard computer displays is valid and shows good agreement with conventional techniques. Scoring digitized images appears to require more time, but this may be outweighed by the considerable advantages of computer storage and retrieval of radiographic images.