Jornal article

Haindl M., Mikeš S., Scarpa G.

Lecture Notes in Computer Science

2007

33-40

0302-9743

prof. Ing. Michal Haindl, DrSc. (ÚTIA - Oddělení rozpoznávání obrazů)

Unsupervised segmentation, mammography, Markov random fields

We present an unsupervised method for fully automatic detection of regions of interest containing fibroglandular tissue in digital screening mammography. The unsupervised segmenter is based on a combination of several unsupervised segmentation results, each in different resolution, using the sum rule. The mammogram tissue textures are locally represented by four causal monospectral random field models recursively evaluated for each pixel. The single-resolution segmentation part of the algorithm is based on the underlying Gaussian mixture model and starts with an over segmented initial estimation which is adaptively modified until the optimal number of homogeneous mammogram segments is reached.