Feasibility study on the application of Raman spectroscopy in the diagnosis of glioma

Glioma is the most common primary intracranial tumor, with a high degree of malignancy. If the tumor can be completely removed during surgery, better treatment results will be achieved. However, it is often difficult to determine the tumor boundary during surgery, so the total tumor resection rate is not high. Raman spectroscopy (RS) is an analytical technique that utilizes the principle of light scattering, RS is an analytical technique, based on the loss of laser energy when the incident laser interacts with the vibrations of molecular chemical bonds. This energy difference corresponds to the frequency of chemical bond vibration, which is also called the fingerprint information of molecular vibration. Therefore, RS can reflect the chemical composition of the sample. This study aims to utilize the characteristic of RS to explore the feasibility of RS in rapid diagnosis of gliomas.

We first conducted RS studies on three glioblastoma cell lines U87, LN229, T98G, and one normal human astrocyte cell line HA1800. Then, we screened out RS bands with significant differences, and validated the selected bands in clinical specimens (three patients with glioblastoma and one contused brain tissue from a patient with severe traumatic brain injury) to verify whether there were similar band intensity differences between cell lines and clinical specimens.

The linear discriminant analysis (LDA) method can completely distinguish the four cell lines. The support vector machine (SVM) algorithm for machine learning was used to build a library model, and it was found that there were significant intensity differences between astrocyte HA1800 and the three glioblastoma cell lines. Subsequently, the same different intensity of RS bands between three clinical glioblastoma specimens and one brain tissue specimen were validated at the same positions using the same algorithm and computational model.

Data collected by Raman microscopy scanning and modeling using SVM algorithm can accurately identify glioma cell lines and ordinary glial cells; and significant differences in RS band intensity between benign and malignant cells can be determined. These differences also exist in clinical specimens of glioblastoma and brain tissue specimens.