The Monte Carlo-based inverse model of diffuse reflectance described in part I of this pair of companion papers was applied to the diffuse reflectance spectra of a set of 17 malignant and 24 normal-benign ex vivo human breast tissue samples. This model allows extraction of physically meaningful tissue parameters, which include the concentration of absorbers and the size and density of scatterers present in tissue. It was assumed that intrinsic absorption could be attributed to oxygenated and deoxygenated hemoglobin and beta-carotene, that scattering could be modeled by spheres of a uniform size distribution, and that the refractive indices of the spheres and the surrounding medium are known. The tissue diffuse reflectance spectra were evaluated over a wavelength range of 400-600 nm. The extracted parameters that showed the statistically most significant differences between malignant and nonmalignant breast tissues were hemoglobin saturation and the mean reduced scattering coefficient. Malignant tissues showed decreased hemoglobin saturation and an increased mean reduced scattering coefficient compared with nonmalignant tissues. A support vector machine classification algorithm was then used to classify a sample as malignant or nonmalignant based on these two extracted parameters and produced a cross-validated sensitivity and specificity of 82% and 92%, respectively.

The features extracted from the Monte Carlo based analysis were hemoglobin saturation, total hemoglobin concentration, beta-carotene concentration and the mean (wavelength averaged) reduced scattering coefficient. Beta-carotene concentration was positively correlated and the mean reduced scattering coefficient was negatively correlated with percent adipose tissue content in normal breast tissues. In addition, there was a statistically significant decrease in the beta-carotene concentration and hemoglobin saturation, and a statistically significant increase in the mean reduced scattering coefficient in malignant tissues compared to non-malignant tissues. The features extracted from the partial least squares analysis were a set of principal components. A subset of principal components showed that the diffuse reflectance spectra of malignant breast tissues displayed an increased intensity over wavelength range of 440-510 nm and a decreased intensity over wavelength range of 510-600 nm, relative to that of non-malignant breast tissues. The diagnostic performance of the classification algorithms based on both feature extraction techniques yielded similar sensitivities and specificities of approximately 80% for discriminating between malignant and non-malignant breast tissues. While both methods yielded similar classification accuracies, the model based approach provided insight into the physiological and structural features that discriminate between malignant and non-malignant breast tissues.

Nonmalignant (n = 36) and malignant (n = 20) tissue samples were obtained from breast cancer and breast reduction surgeries. These tissues were characterized using multiple excitation wavelength fluorescence spectroscopy and diffuse reflectance spectroscopy in the ultraviolet-visible wavelength range, immediately after excision. Spectra were then analyzed using principal component analysis (PCA) as a data reduction technique. PCA was performed on each fluorescence spectrum, as well as on the diffuse reflectance spectrum individually, to establish a set of principal components for each spectrum. A Wilcoxon rank-sum test was used to determine which principal components show statistically significant differences between malignant and nonmalignant tissues. Finally, a support vector machine (SVM) algorithm was utilized to classify the samples based on the diagnostically useful principal components. Cross-validation of this nonparametric algorithm was carried out to determine its classification accuracy in an unbiased manner. Multiexcitation fluorescence spectroscopy was successful in discriminating malignant and nonmalignant tissues, with a sensitivity and specificity of 70% and 92%, respectively. The sensitivity (30%) and specificity (78%) of diffuse reflectance spectroscopy alone was significantly lower. Combining fluorescence and diffuse reflectance spectra did not improve the classification accuracy of an algorithm based on fluorescence spectra alone. The fluorescence excitation-emission wavelengths identified as being diagnostic from the PCA-SVM algorithm suggest that the important fluorophores for breast cancer diagnosis are most likely tryptophan, NAD(P)H and flavoproteins.

We explore the use of Monte-Carlo-model-based approaches for the analysis of fluorescence and diffuse reflectance spectra measured ex vivo from breast tissues. These models are used to extract the absorption, scattering, and fluorescence properties of malignant and nonmalignant tissues and to diagnose breast cancer based on these intrinsic tissue properties. Absorption and scattering properties, including β-carotene concentration, total hemoglobin concentration, hemoglobin saturation, and the mean reduced scattering coefficient are derived from diffuse reflectance spectra using a previously developed Monte Carlo model of diffuse reflectance. A Monte Carlo model of fluorescence described in an earlier manuscript was employed to retrieve the intrinsic fluorescence spectra. The intrinsic fluorescence spectra were decomposed into several contributing components, which we attribute to endogenous fluorophores that may present in breast tissues including collagen, NADH, and retinol/ vitamin A. The model-based approaches removes any dependency on the instrument and probe geometry. The relative fluorescence contributions of individual fluorescing components, as well as β-carotene concentration, hemoglobin saturation, and the mean reduced scattering coefficient display statistically significant differences between malignant and adipose breast tissues. The hemoglobin saturation and the reduced scattering coefficient display statistically significant differences between malignant and fibrous/benign breast tissues. A linear support vector machine classification using (1) fluorescence properties alone, (2) absorption and scattering properties alone, and (3) the combination of all tissue properties achieves comparable classification accuracies of 81 to 84% in sensitivity and 75 to 89% in specificity for discriminating malignant from nonmalignant breast tissues, suggesting each set of tissue properties are diagnostically useful for the discrimination of breast malignancy.

The goal of the work is to experimentally verify Monte Carlo modeling of fluorescence and diffuse reflectance measurements in turbid, tissue phantom models. In particular, two series of simulations and experiments, in which one optical parameter (absorption or scattering coefficient) is varied while the other is fixed, are carried out to assess the effect of the absorption coefficient (mu(a)) and scattering coefficient (mu(s)) on the fluorescence and diffuse reflectance measured from a turbid medium. Moreover, simulations and experiments are carried out for several fiber optic probe geometries that are designed to sample small tissue volumes. Additionally, a group of conversion expressions are derived to convert the optical properties and fluorescence quantum yield measured from tissue phantoms for use in Monte Carlo simulations. The conversions account for the differences between the definitions of the absorption coefficient and fluorescence quantum yield of fluorophores in a tissue phantom model and those in a Monte Carlo simulation. The results indicate that there is good agreement between the simulated and experimentally measured results in most cases. This dataset can serve as a systematic validation of Monte Carlo modeling of fluorescent light propagation in tissues. The simulations are carried out for a wide range of absorption and scattering coefficients as well as ratios of scattering coefficient to absorption coefficient, and thus would be applicable to tissue optical properties over a wide wavelength range (UV-visible/near infrared). The fiber optic probe geometries that are modeled in this study include those commonly used for measuring fluorescence from tissues in practice.