Conventional versus digital mammography in the analysis of screen-detected lesions with low positive predictive value

Purpose: To compare the performance of screen-film and digital mammography in the assessment of screen-detected breast lesions. Materials and methods: A series of 100 consecutive mammographic screen-detected lesions (65 masses, 6 architectural distortions, 29 microcalcifications) deserving diagnostic assessment and judged to have a low positive predictive value underwent screen-film mammography (SFM) and digital mammography by a Fuji computed radiography system (FCR) (double exposure, same view, without removing compression) of the corresponding breast. Three sets of images (SFM, hard copy and soft copy FCR) were read, blind of assessment outcome, by three experienced radiologists. For the three different imaging modalities a contrast-detail analysis, dose evaluation and diagnostic accuracy by means of ROC analysis were performed. At the end of the diagnostic workup all suspicious cases (20) underwent surgical biopsy and were histologically confirmed as cancers and the cases which were negative or benign at assessment (80) were followed up for a period of 12-20 months. During the follow-up period two more cases proved to be cancers at subsequent examinations. Results: Contrast-detail analysis gives better image quality for FCR compared to SFM at the same delivered dose, whilst in ROC analysis the SFM (AUC 0.7158), hard copy FCR (AUC 0.7404) and soft copy FCR (AUC 0.7501) (chi(2) = 1.30, p = 0.5220) are equivalent. Conclusion: FCR has a diagnostic performance equivalent to SFM in the assessment of screen-detected lesions with a low positive predictive value for cancer and it may be safely included in routine screening practice. (C) 2004 Elsevier Ireland Ltd. All rights reserved.