Abstract: Background and purpose: Full-field digital mammography (FFDM) is not only a routine examination of breast, but also the preferred method of breast cancer screening recognized at home and abroad. But breast tissue is sensitive to X-ray, and exposing to ionizing radiation can raise the risk of cancer. This paper aimed to discuss the relationship among the radiation dose of FFDM, breast density and body composition. Methods: A total of 104 breast cancer patients were retrospectively analyzed with breast mammography and bone density examination in Fudan University Shanghai Cancer Center from Jan. 2016 to Dec. 2016. Then we collected the breast image data on breast density, compression thickness, radiation dose, signal-to-noise ratio (SNR), and so on. We recorded the weight, height, body mass index (BMI), body surface area, spine and thigh muscle percentage at the same time, and analyzed the tube current, tube voltage, radiation dose, compression thickness and SNR at two different projection positions of each type of breast density. Each type of breast density with weight, BMI, spine and thigh muscle percentage and subjective image quality were analyzed. The correlation between physical indexes and radiation dose was further analyzed. Results: In all of the 104 subjects, there were 17 cases of type a breast density, 31 cases of type b, 40 cases of type c and 16 cases of type d. There were significant diferences in tube current, tube voltage and SNR in carnio-caudal (CC) view between the groups (P<0.05), especially in patients with type c breast density whose SNR was 3.75 + 0.96, significantly lower than that of type a (5.21+1.36), type b (4.81+1.09) and type d (4.71+1.34) patients. There were significant differences in tube voltage and compression thickness in medial lateral oblique (MLO) view between the groups with various types of breast density (P<0.05). The SNR of type c breast density was the lowest among all types of density, which was close to statistical significance. There were significant differences in physical indexes of each type of breast density, including weight, body surface area, BMI, spine and thigh muscle percentage (P<0.05), and with the increase of breast density, the weight, body surface area and BMI showed a decreasing trend, while the spine and thigh muscle percentage increased. Correlation analysis showed that radiation dose and thigh muscle percentage had the highest correlation in CC view with a moderate negative correlation (r =-0.311). The radiation dose and BMI had the highest correlation in MLO view (r=0.492), followed by the correlation of radiation dose and thigh muscle percentage (r=-0.418), and both showed a moderate correlation. Conclusion: Patients with type c breast density have the highest radiation dose, the lowest SNR and relatively low image quality. The thigh muscle percentage is closely related to breast density and radiation dose, which opens up a new research direction for radiation protection in the future.

Key words: Mammography, Muscle percentage, Radiation dose, Breast density, Body mass index, Signal-noise ratio