Autofluorescence of Breast Tissues: Evaluation of Discriminating Algorithms for Diagnosis of Normal, Benign, and Malignant Conditions

Abstract

Objective: We evaluated different discriminating algorithms for classifying laser-induced fluorescence spectra of normal, benign, and malignant breast tissues that were obtained with 325-nm excitation. Background Data: Mammography and histopathology are the conventional gold standard methods of screening and diagnosis of breast cancers, respectively. The former is prone to a high rate of false-positive results and poses the risk of repeated exposure to ionizing radiation, whereas the latter suffers from subjective interpretations of morphological features. Thus the development of a more reliable detection and screening methodology is of great interest to those practicing breast cancer management. Several studies have demonstrated the efficacy of optical spectroscopy in diagnosing cancer and other biomedical applications. Materials and Methods: Autofluorescence spectra of normal, benign, and malignant breast tissues, with 325-nm excitation, were recorded. The data were subjected to diverse discriminating algorithms ranging from intensities and ratios of curve-resolved bands to principal components analysis (PCA)-derived parameters. Results: Intensity plots of collagen and NADPH, two known fluorescent biomarkers, yielded accurate classification of the different tissue types. PCA was carried out on both unsupervised and supervised methods, and both approaches yielded accurate classification. In the case of the supervised classification, the developed standard sets were verified and evaluated. The limit test approach provided unambiguous and objective classification, and this method also has the advantage of being user-friendly, so untrained personnel can directly compare unknown spectra against standard sets to make diagnoses instantly, objectively, and unambiguously. Conclusion: The results obtained in this study further support the efficacy of 325-nm-induced autofluorescence, and demonstrate the suitability of limit test analysis as a means of objectively and unambiguously classifying breast tissues.