Optical Screening System Using Biospeckle Imaging and Texture-Based Analysis for Breast Cancer Detection

Breast cancer (BC) is a prevalent cause of cancerrelated mortality in women, posing a significant public health challenge that demands effective screening techniques for cancer detection. To address this need, biospeckle imaging, as a non-destructive and cost-efficient method, has emerged to assess the functionality and activity of biological tissues. However, it is necessary to investigate algorithms capable of interpreting and analyzing biospeckle patterns susceptible by experimental conditions. In this investigation, a novel optical screening system utilizing biospeckle imaging for breast cancer screening is proposed, incorporating a promising texture-based multi-neighborhood local entropy analysis. The biospeckle patterns are generated from samples of human breast tissue. The analysis approach extracts histogram-based features from texture maps to capture spatial intensity and phase variations of biospeckle patterns associated with breast tissue functionality. Canonical correlation analysis (CCA) is employed to identify the most informative features. By employing a nonlinear support vector machine (SVM) algorithm, the relationship between the selected features and breast conditions is established. The proposed method introduces a viable imaging system for BC prediction, achieving a maximum accuracy and recall of $97.29 \%$, precision of $98.05 \%$, and $F 1$ score of $97.98 \%$.