Non-invasive diffused reflected/transmitted signature accompanied with hyperspectral imaging for breast cancer early diagnosis

Abstract

IntroductionBreast cancer (BC) is a prevalent form of cancer worldwide and a leading cause of cancer-related deaths. There is an urgent need for the development of new methods to detect, diagnose, and monitor BC at an early stage. Light propagation and interaction with molecular species, could be utilized for diagnostic purposes. Incident light reflection and transmission are influenced by the changes of cell structure and dielectric properties. Any variations in these components could impact the reflected and transmitted signals, offering potential diagnostic applications. MethodsIn this study, non-invasive, polychromatic light source over the visible range (400–900 nm) was employed to illuminate controlled normal and BC tissue samples. The resulted diffused reflected and transmitted spectra were captured and analyzed using hyperspectral (HS) camera and designed cross-correlation algorithm. ResultsDiffused reflected signature demonstrated significant difference between normal and BC tissues over the band 500–700 nm. At 600 nm, normal tissue demonstrated diffused reflection intensity of 2304 relative to 764 for malignant tissues. Transmitted signature exhibited good discrimination capabilities over the band 600–900 nm. At 750 nm, normal tissue demonstrated transmission signal intensity of 34 relative to 21 for malignant tissues.Results were validated via generated HS 2D images, for diffused reflected and transmitted signals. Noise reduction, using moving average filter (K = 10), was adopted to enhance HS images. Contrast was improved through de-convolution with reference. Magnitude was calculated using fast Fourier transform for the concated diffused reflected and transmitted HS images. Additionally, phase shift analysis was performed to assess the differences in arrival time. ConclusionThe optimal wavelength to distinguish between normal and malignant breast tissues was 600 nm for the diffused reflected HS images; Furthermore, the transmitted HS images at 750 nm demonstrated high contrast and specificity. The developed HS imaging technique secured high levels of accuracy, specificity, and sensitivity for diffused reflected and transmitted images at 600 nm and 750 nm, respectively; K-means clustering algorithm offered precise outline and delineation of malignant areas at these wavelengths.