Recognizing Architectural Distortion in Mammogram: A Multiscale Texture Modeling Approach with GMM

Abstract

We propose a generative model for constructing an efficient set of distinctive textures for recognizing architectural distortion in digital mammograms. In the first layer of the proposed two-layer architecture, the mammogram is analyzed by a multiscale oriented filter bank to form texture descriptor of vectorized filter responses. Our model presumes that every mammogram can be characterized by a "bag of primitive texture patterns" and the set of textural primitives (or textons) is represented by a mixture of Gaussians which builds up the second layer of the proposed model. The observed textural descriptor in the first layer is assumed to be a stochastic realization of one (hard mapping) or more (soft mapping) textural primitive(s) from the second layer. The results obtained on two publicly available datasets, namely Mammographic Image Analysis Society (MIAS) and Digital Database for Screening Mammography (DDSM), demonstrate the efficacy of the proposed approach.

A textural approach for recognizing architectural distortion in mammograms

“…To demonstrate the effectiveness of our proposed approach, the results were compared with one of the most recent works [7]. Biswas et al [7] present a two-layer generative framework to construct an impressive set of distinctive texture for recognition of architectural distortion.…”

“…Biswas et al [7] present a two-layer generative framework to construct an impressive set of distinctive texture for recognition of architectural distortion. In their first layer, Multi-Scale Rotationally Invariant Derivatives of Gaussians (MSRI-DOG) filters are used to form texture descriptor of filter response.…”

“…The method proposed in [6], utilizes Gabor filters and phase portrait analysis to determine the potential sites of architectural distortion in prior mammograms. Biswas et al [7] utilized a mixture of Gaussian distributions in their two-layer generative method to model variety of mammogram texture. However, significant enhancement in the performance of CAD systems is still needed for recognizing architectural distortion.…”

Computational unfoldment of mammograms

“…The reduced feature set was fed to ANN to classify the images. Biswas et al [24] in their model characterized every mammogram by textural patterns, which were represented by a mixture of Gaussians. Multiscale oriented filters were used to obtain texture maps.…”

Haralick fetaures based mammogram classification system

“…A general methodology where different types of mammogram textures can be modeled by a mixture of Gaussian distributions is presented in [3]. The presented two layer architecture first collects the low level rotation invariant textural features at different scales and then learns latent textural primitives from the collected features by Gaussian mixture model.…”