Diagnostic Evaluation of CR Mammography Using Three Different Image-Processing Patterns

Abstract

Diagnostic evaluation of computed radiography (CR) was performed in 119 cases of mammary cancer, using the following three different image-processing patterns : 1) the xero-pattern, in which high-frequency components were emphasized and xero-radiogram-like images were obtained, 2) the conventional pattern, in which high-frequency components were attenuated and conventional radiogram-like images were obtained, and 3) the intermediate pattern, in which images intermediate between 1) and 2) were obtained. In performing CR mammography using these three patterns in the same patients, the incidences of detection of tumors, calcifications, spiculae and trabecular abnormalities were examined and compared. With regard to tumor detection, the xero-pattern was found to be useful for clarifying the tumor border due to the edge effect, although the tumor diameter was reduced by 7.1% on average as compared with that examined by the conventional pattern. Such reduction was more marked in smaller tumors. As for detection of calcifications, higher incidences were obtained when patterns, emphasizing the edge effect were used. By contrast, for detection of spiculae and trabecular abnormalities, the incidences were lower when patterns, emphasizing the edge effect were used. Exceptionally, the incidence of detecting spiculae in cases of scirrhous carcinoma was the same for all three image -processing patterns. In conclusion, CR mammograhy using the conventional pattern seems the most appropriate method for mammary cancer mass-screening.