The aim of this study was to analyze interobserver agreement in evaluation of breast composition and differences in sensitivity according to breast composition. ［Materials and methods］In 51cases of breast cancer, which were confirmed either by operation or needle biopsy between April2011and March2012at Tsukuba Medical Center Hospital, pre- treatment two-view mammograms were available. Ninety-nine normal or benign cases were also chosen. Five qualified observers who had no information about the cases used two 5M pixel monitors and evaluated all１ 50cases in terms of breast category and breast composition using four different classifications: the classification usually used by the observer, the classification prescribed by the Japanese mammography guidelines, a classification that takes into consideration whether ultrasound examination should be added, and the classification based on the BI-RADS classification, which assesses the area of fibro-glandular tissue. ［Results］Interobserver agreement was higher for the classification prescribed by the Japanese mammography guidelines and the classification usually employed by the observer, compared with the other two classifications. The sensitivities were61.5～67.3％ for the cases overall and69.6～81.2％ for cases involving fatty breasts or breasts with scattered glands. The sensitivity for women aged50years or over was71.4～77.1％, regardless of breast composition. ［Conclusion］Interobserver agreement was high for breast composition as defined by the Japanese guidelines. For fatty breasts or breasts with scattered glands, the breast cancer sensitivity was almost the same as that for women aged 50years or over.
To ascertain the current status of soft―copy mammography and its interpretation, a questionnaire was sent to 415 institutions in the Kyushu bloc, of which 275 institutions（66％）returned the questionnaire. The proportion of institutions that had fully or partially installed the equipment, or which planned to do so, was 51％. The proportion exceeded 60％ if institutions that planned to install the equipment in the near future were included. We investigated 144 sets of digital mammography equipment at 120 institutions, among which CR（computed radiography）and DR（digital radiography; flat panel detector system） accounted for 61％ and 37％, respectively. The percentage use of 5M pixel liquid crystal displays when radiological technologists were checking patients’ precise positioning and for performing additional spot compressionwas only 42％ and 46％, respectively. On the other hand, the rate was 81％ for interpretation by the doctor. The rate of periodic quality control（QC）of the display and statistical management of diagnostic reports was low. It was suggested that evaluation of image quality, the interpretation environment, and the contents of inspection were insufficient. Although the training course in digital mammography QC and certification and license in soft ―copy mammography stipulated by the Japan Central Institution on Quality Assurance of Breast Cancer Screening should serve as an aid to information acquisition, it appears to be insufficiently functional. Therefore, information about accuracy control needs to be shared in the Kyushu bloc.
Clinical breast examination（CBE）including inspection and palpation, mammography（MMG） and ultrasonography（US）have mainly been used for breast cancer screening, and each have different rates of cancer detection or undetectability according to their specific characteristics. We tried to search for a combination of examination methods that would facilitate effective breast cancer detection at screening, and the possibility of excluding CBE for routine screening was discussed. During a period of 2 years and 10 months, from January 2009 through October 2012, a total of 16,112women underwent screening at our institutions and 44 were found to have breast cancers. Their records at the time of initial screening, detailed examination and surgical operation were reviewed. The cancer detection rates were 40.9％ for CBE only, 90.9％ for mammography only, 97.7％ for US only, 100％ for CBE and MMG, 97.7％ for CBE and US, and 100％ for MMG and US. With regard to the stage of cancers at the time of detection, 0％ detected by CBE only were Stage0, which was a lower rate of detection than that for Stage I cancers by CBE only compared with those detected by other single examination methods, and the detection rate for Stage II cancers by any of the three methods was the same. The cancers not detected by CBE were those with calcification（which were detected by MMG）and small cancers less than 2cm in diameter（which were detected by US）. Cancers not detected by MMG were those in young individuals, those located in a dense breast mass, and small cancers less than 1.5cm in diameter（which were detected by US）. Cancers undetectable by US were those with calcification（which were detected by MMG） . From the above results, a combination of MMG and US appears to yield the best cancer detection rate. Stage analysis of detected cancers revealed that CBE was the least effective examination method. A combination of MMG and US proved to be effective for mutually detecting specific types of cancers that are difficult to visualize by any one of the two methods alone. CBE may be omitted from mass screening provided that screenees having symptoms are excluded by initially treating them as patients.