# Model-free classification of X-ray scattering signals applied to image segmentation

Lutz-Bueno, V; Arboleda, C; Leu, L; Blunt, M J; Busch, A; Georgiadis, A; Bertier, P; Schmatz, J; Varga, Z; Villanueva-Perez, P; Wang, Z; Lebugle, M; David, C; Stampanoni, M; Diaz, A; Guizar-Sicairos, M; Menzel, A (2018). Model-free classification of X-ray scattering signals applied to image segmentation. Journal of Applied Crystallography, 51(5):1378-1386.

## Abstract

In most cases, the analysis of small-angle and wide-angle X-ray scattering (SAXS and WAXS, respectively) requires a theoretical model to describe the sample's scattering, complicating the interpretation of the scattering resulting from complex heterogeneous samples. This is the reason why, in general, the analysis of a large number of scattering patterns, such as are generated by time-resolved and scanning methods, remains challenging. Here, a model-free classification method to separate SAXS/WAXS signals on the basis of their inflection points is introduced and demonstrated. This article focuses on the segmentation of scanning SAXS/WAXS maps for which each pixel corresponds to an azimuthally integrated scattering curve. In such a way, the sample composition distribution can be segmented through signal classification without applying a model or previous sample knowledge. Dimensionality reduction and clustering algorithms are employed to classify SAXS/WAXS signals according to their similarity. The number of clusters, <jats:italic>i.e.</jats:italic> the main sample regions detected by SAXS/WAXS signal similarity, is automatically estimated. From each cluster, a main representative SAXS/WAXS signal is extracted to uncover the spatial distribution of the mixtures of phases that form the sample. As examples of applications, a mudrock sample and two breast tissue lesions are segmented.

## Abstract

In most cases, the analysis of small-angle and wide-angle X-ray scattering (SAXS and WAXS, respectively) requires a theoretical model to describe the sample's scattering, complicating the interpretation of the scattering resulting from complex heterogeneous samples. This is the reason why, in general, the analysis of a large number of scattering patterns, such as are generated by time-resolved and scanning methods, remains challenging. Here, a model-free classification method to separate SAXS/WAXS signals on the basis of their inflection points is introduced and demonstrated. This article focuses on the segmentation of scanning SAXS/WAXS maps for which each pixel corresponds to an azimuthally integrated scattering curve. In such a way, the sample composition distribution can be segmented through signal classification without applying a model or previous sample knowledge. Dimensionality reduction and clustering algorithms are employed to classify SAXS/WAXS signals according to their similarity. The number of clusters, <jats:italic>i.e.</jats:italic> the main sample regions detected by SAXS/WAXS signal similarity, is automatically estimated. From each cluster, a main representative SAXS/WAXS signal is extracted to uncover the spatial distribution of the mixtures of phases that form the sample. As examples of applications, a mudrock sample and two breast tissue lesions are segmented.

## Statistics

### Citations

Dimensions.ai Metrics
6 citations in Web of Science®
5 citations in Scopus®

### Altmetrics

24 downloads since deposited on 23 Nov 2018