Document Type : Methodological papers
Authors
1
Shizuoka Cancer Center Hospital, Japan.
2
Department of Breast Imaging and Breast Interventional Radiology, Shizuoka Cancer Center Hospital, Shizuoka, Japan.
3
Department of Diagnostic Radiology, Shizuoka Cancer Center Hospital, Shizuoka, Japan.
4
Department of Radiology, Hamamatsu University School of Medicine, Shizuoka, Japan.
5
Department of Radiology, Fujieda Municipal General Hospital, Shizuoka, Japan.
6
Clinical Research Center, Shizuoka Cancer Center Hospital, Shizuoka, Japan.
Abstract
Purpose: Breast density assessment is an essential component of risk-stratified and supplemental breast cancer screenings for the future generation. International collaborative research and cooperation are required to determine global trends and differences in mammographic breast density. Due to the cost and energy associated with the use of film or digital mammograms, international breast density assessment studies tend to be challenging. A novel breast density assessment method using portable document format (PDF) is currently explored and the justification is needed. This study aimed to investigate the inter-method agreement between two viewing platforms – mammograms on PDF and mammograms on a workstation – for a subjective breast density assessment. Methods: Three radiologists assessed 100 cases with 200 normal bilateral mediolateral oblique view images twice at 1-month interval using mammograms on PDF and mammograms on a workstation. Further, to assess intra-reader agreement, one reader performed two readings of the PDF set at 1-month interval. The reading order of the images was changed each time. Weighted kappa coefficient (κw) was used to assess inter-method and intra-observer agreements. Results: The average inter-method agreement was substantial (κw = 0.74, range = 0.73–0.76) on a four-grade scale (fatty, scattered, heterogeneously dense, or extremely dense) and almost perfect (κw = 0.81, range = 0.78–0.84) on a two-grade scale (non-dense or dense). The intra-observer agreement was almost perfect on a four-grade scale (κw = 0.85) and two-grade scale (κw = 0.90). Conclusion: The PDF and mammography workstation readings for breast density assessment had almost perfect inter-method agreements. Moreover, the intra-observer agreement of PDF reading was almost perfect.
Keywords
Main Subjects