Identification of breast vascular calcium deposition in digital mammography by linear structure analysis

Computerized detection of vascular calcium depositions in mamagraphy is a new research topics, which is driven by the clinical hypothesis of the association with many related cardiovascular diseases. In several previous studies [7, 9], calcification cue plays a very important role in the computerized analysis. We observe that vascular calcium depositions can be identified with high confidence if they appear in a bright railway pattern. Accordingly, a linear structure analysis method is introduced in this study to detect most true calcifications and also keep the false positives as little as possible. The proposed method is tested with 40 mammograms and achieves performance of 93.8±1.3% in sensitivity and 84.7±3.9% in specificity. The output of this linear structure analysis may provide more reliable calcification cue for the subsequent vessel tracking process, which will be investigated in the future.