Discrimination of normal and malignant mouse ovarian surface epithelial cells in vitro using Raman microspectroscopy

Abstract

Raman microspectroscopy in conjunction with multivariate statistical analysis is a powerful technique for label-free classification of live cells based on their molecular composition, which can be correlated to variations in protein, DNA/RNA, and lipid macromolecules. We apply this technique in vitro, to discriminate between normal mouse ovarian surface epithelial (MOSE) cells and spontaneously-transformed ovarian surface epithelial (STOSE) cells, which are derived from the MOSE cells and are a model for high-grade serous ovarian cancer. The Raman spectra collected from individual cells undergo initial preprocessing (background subtraction, normalization and noise reduction) to yield true Raman spectra representative of the cells for subsequent statistical analysis. Using Principal Component Analysis (PCA) followed by Linear Discriminant Analysis (LDA) yields a separation of the cells into the two groups (MOSE and STOSE). This classification model has a sensitivity and specificity of 92% and 85%, respectively, after treatment of the cells to bring them into cell cycle synchrony. The main source of this separation is correlated with the increased nucleic acid content in the malignant OSE cells. As expected, a lower accuracy of 72% is obtained with asynchronous MOSE and STOSE cell populations. These results are expected to have a positive impact on the future development of improved strategies for early detection and therapeutics related to ovarian cancer.