A new idea for visualization of lesions distribution in mammogram based on CPD registration method

Mammography is currently the most effective technique for breast cancer. Lesions distribution can provide support for clinical diagnosis and epidemiological studies.

We presented a new idea to help radiologists study breast lesions distribution conveniently. We also developed an automatic tool based on this idea which could show visualization of lesions distribution in a standard mammogram.

Firstly, establishing a lesion database to study; then, extracting breast contours and match different women's mammograms to a standard mammogram; finally, showing the lesion distribution in the standard mammogram, and providing the distribution statistics. The crucial process of developing this tool was matching different women's mammograms correctly. We used a hybrid breast contour extraction method combined with coherent point drift method to match different women's mammograms.

We tested our automatic tool by four mass datasets of 641 images. The distribution results shown by the tool were consistent with the results counted according to their reports and mammograms by manual. We also discussed the registration error that was less than 3.3 mm in average distance.

The new idea is effective and the automatic tool can provide lesions distribution results which are consistent with radiologists simply and conveniently.