Combining Adaptive Segmentation Approach for Improving Multi-Resolution Image Registration on X-Ray Mammograms Acquired using Fischer's Fused Full Field Digital Mammography and Ultrasound System (FFDMUS)

J. Suri, Y. Guo, Y. Sun, T. Danielson, and R. Janer (USA)

Keywords

Fused system, image registration, adaptive segmentation, mutual information, polyline distance

Abstract

: Mammogram registration can help avoid biopsies and help follow-up analysis. Mammogram registration is a challenge because of the compressibility of the breast tissue. We used the CIRS breast phantom image to study the registration methodology. The phantom images were acquired in Fused Full Field Digital Mammogram and Ultrasound (FFDMUS) framework. The acquired images were registered using our proposed three-stage registration system. The registration results were measured and evaluated, and the comparison was made for the registration procedure with and without mask operation. It is demonstrated that the mask operation can help improve the performance of the intensity based registration. The mean error from the polyline distance measurement is improved by 19.6%, and the standard deviation of the error is improved by 13.7%.

Important Links:

Go Back