Sherif Abbas, Nihal Simsek Ozek, Salih Emri, Deniz Koksal, Mete Severcan, Feride Severcan
Journal of Biomedical Optics, Vol. 23, Issue 10, 105003, (October 2018) https://doi.org/10.1117/1.JBO.23.10.105003
TOPICS: Principal component analysis, Proteins, Mesothelioma, Spectroscopy, Chemometrics, Diagnostics, Statistical analysis, FT-IR spectroscopy, Statistical modeling, Fourier spectroscopy
This study was conducted to differentiate malignant pleural mesothelioma (MPM) from lung cancer (LC) and benign pleural effusion (BPE) from pleural fluids using the diagnostic power of Fourier transform-infrared spectroscopy with attenuated total reflectance mode coupled with chemometrics. Infrared spectra of MPM (n = 24), LC (n = 20), and BPE (n = 25) were collected, and hierarchical cluster analysis (HCA) and principal component analysis (PCA) were applied to their spectra. HCA results indicated that MPM was differentiated from LC with 100% sensitivity and 100% specificity and from BPE, with 100% sensitivity and 88% specificity, which were also confirmed by PCA score plots. PCA loading plots indicated that these separations originated mainly from lipids, proteins, and nucleic acids-related spectral bands. There was significantly higher lipid, protein, nucleic acid, and glucose contents in the MPM and LC. However, the significant changes in triglyceride and cholesterol ester content, protein and nucleic acid structure, a lower membrane fluidity, and higher membrane order were only observed in the MPM. To check the classification success of some test samples/each group, soft independent modeling of class analogies was performed and 96.2% overall classification success was obtained. This approach can provide a rapid and inexpensive methodology for the efficient differentiation of MPM from other pleural effusions.