Label‐free serum detection based on Raman spectroscopy for the diagnosis and classification of glioma

Glioma is the most prevalent malignant cancer in the central nervous system and can cause significant mortality and morbidity. A rapid, convenient, accurate, and relatively noninvasive diagnostic method for glioma is important and urgently needed. In this study, we investigated the feasibility of using Raman spectroscopy to discriminate patients with glioma from healthy individuals. Serum samples were collected from healthy individuals (n = 86) and patients with glioma [high‐grade glioma (HGG) n = 75, low‐grade glioma (LGG) n = 60]. All spectra were collected with a 785‐nm wavelength laser in the range of 400–1800 cm−1. A total of three spectra were recorded for each sample, and every spectrum was integrated for 12 s and averaged over five accumulations. Principal component analysis and linear discriminant analysis models were combined to classify the Raman spectra of different groups. The correct classification ratios were 95.35, 93.33, and 93.34% for the normal, HGG, and LGG groups, respectively, and the total accuracy was 94.12%. The sensitivity, specificity, and accuracy of differentiating the HGG group from the normal group were 96.00, 96.51, and 96.27%, respectively, with an area under the curve of 0.997; in addition, the sensitivity, specificity, and accuracy of differentiating the LGG group from the normal group were 96.67%, 98.84%, and 97.95%, respectively, with an area under the curve of 0.999. Our study results suggested that the rapid and noninvasive detection method based on principal component analysis and linear discriminant analysis combined with Raman spectroscopy is a highly promising tool for the early diagnosis of glioma. In this study, we investigated the feasibility of using Raman spectroscopy combining with principal component analysis (PCA) and linear discriminant analysis (LDA) to discriminate glioma patients from normal people using serum, which displayed a good discrimination effect except for a small overlap.