Objective: To assess the efficacy of conventional Raman spectroscopy in combination with discriminating parameters, Mahalanobis distance, spectral residuals, and "limit test" methodology in differentiation of normal and malignant ovarian tissues. Background Data: Ovarian cancer is the second most common cancer among women and the leading cause of death among gynecologic malignancies. Initial laparotomy and subsequent frozen section analysis can influence the surgical management of ovarian cancers. Although frozen section pathology is sensitive and specific enough, interpretation is often subjective, time consuming, and requires highly skilled personnel. Raman spectroscopy is sensitive to biochemical variations in the samples, rapid, more objective, and amenable to multivariate statistical tools. It can therefore be an ideal tool for discrimination between normal and malignant ovarian tissues. Methods: 72 Spectra from eight normal and seven malignant ovarian tissues were recorded by conventional near-infrared (NIR) Raman spectroscopy (excitation wavelength of 785 nm). Spectral data were analyzed by principal components analysis (PCA) and other discriminating parameters such as Mahalanobis distance, spectral residuals, and a multiparametric limit test approach. Results: A mean malignant spectrum exhibits a broader amide I band, a stronger amide III band, a minor blue shift in the δCH2 band, and a hump around 1480 cm -1 compared to a normal spectrum. The normal spectra show relatively stronger peaks around the 855 and 940 cm-1 region. Scores of factor 1 as well as Mahalanobis distance and spectral residuals gave good classification among the tissue types. The limit test approach provided unambiguous and objective discrimination. Conclusion: The findings of this study demonstrate the efficacy of conventional Raman spectroscopy and our statistical methodologies in discrimination of normal from malignant ovarian tissues. Prospectively, by evaluating the models and developing suitable fiberoptic probes, this technique could be useful in diagnosis during initial laparotomy.
All Science Journal Classification (ASJC) codes
- Biomedical Engineering
- Radiology Nuclear Medicine and imaging