Automatic detection of architectural distortion on mammograms

the main objective, is to characterize the radial spiculae of a typical architectural distortion.

Breast cancer is the most frequently diagnosed cancer in women, representing a 16% of all female cancers. However, breast cancer is fully treatable when detected in early stages and mammography is yet the more effective method for detecting early abnormalities, even though some studies have reported that between 10% and 25% of breast cancers are not detected. Such variability is present because mammographic examination is particularly dicult, especially when radiological signs are hidden by the complex superposition of soft tissues. Ultimately, development of Computer Assisted Diagnosis Systems (CAD) in mammography has proven to be eective in finding calcications and masses. Nevertheless, they have been found to fail in detecting architectural distortion with adequate levels of accuracy, typically of 49 %. An architectural distortion is any change of the spatial distribution of the breast connective tissue, in general with no visible mass. The typical retraction of tissues is variable and gradual, sometimes showing spiculations radiating from a point, sometimes accompanied of a sort of focal retraction and at some point, including also distortion at the parenchyma boundary. The set of radiological signs of an architectural distortion has been collected and formalized within the Breast Imaging Reporting and Database Systems (BI-RADS) standard. Among the dierent lesions directly related with the presence of breast cancer, the architectural distortion (AD) has been reported as the most commonly missed abnormality in false-negative cases. the main objective of this work, is to characterize the radial spiculae of a typical architectural distortion to improve diagnosis sensitivity.