Article
Delineation and identification of primary and secondary brain tumours by multiple analyses of Raman spectroscopic datasets
Abgrenzung und Identifizierung von primären und sekundären Hirntumoren durch multiple Analyse Raman-spektroskopischer Datensätze
Search Medline for
Authors
-
Ortrud Uckermann
- Universitätsklinikum Carl Gustav Carus Dresden, Klinik und Poliklinik für Neurochirurgie, Dresden, Deutschland
-
Roberta Galli
- Technische Universität Dresden, Klinisches Sensoring und Monitoring, Dresden, Deutschland
-
Edmund Koch
- Technische Universität Dresden, Klinisches Sensoring und Monitoring, Dresden, Deutschland
-
Gabriele Schackert
- Universitätsklinikum Carl Gustav Carus Dresden, Klinik und Poliklinik für Neurochirurgie, Dresden, Deutschland
-
Gerald Steiner
- Technische Universität Dresden, Klinisches Sensoring und Monitoring, Dresden, Deutschland
-
Matthias Kirsch
- Universitätsklinikum Carl Gustav Carus Dresden, Klinik und Poliklinik für Neurochirurgie, Dresden, Deutschland
| Published: | June 26, 2020 |
|---|
Outline
Text
Objective: Raman spectroscopy provides a comprehensive molecular signature of the tissue and has been suggested as innovative tool for intraoperative brain tumor analysis. Here, we performed multiple analysis of the Raman spectroscopic dataset with the aim to extract different diagnostic tumor parameters.
Methods: Fresh, unprocessed biopsies of 202 tumor patients were investigated. Non-neoplastic brain tissue was obtained during surgical treatment of drug-resistant epilepsy (n=7). Within 30 min from tissue resection, five Raman spectra were acquired on each sample using a standard Raman microscope and 785 nm laser excitation. After baseline correction and normalization, principal component analysis (PCA) of Raman spectra was applied. Quadratic discriminant analysis of PCA scores was performed in reference to the diagnostic information retrieved by standard histopathology.
Results: Raman bands of lipids at 1090, 1297, 1438, 2708 and 2847 cm-1 were significantly reduced in primary and secondary brain tumors compared to non-tumor brain. In contrast, Raman bands assigned to proteins at 1003, 1240, 1660, 2945 cm-1 were increased. Spectral changes were more pronounced in high grade glioma than in low grade glioma which is in line with previous studies. Classification led to the correct recognition of all non-neoplastic biopsies and of 95% of the investigated tumor biopsies, confirming the potential of Raman spectroscopy for tumor delineation. Further analysis of the Raman signatures of the tumor samples allowed the characterization of the tumor type: Differentiation of glioma and brain metastases was obtained with an accuracy of 90%. Moreover, oligodendroglioma and IDH1-mutant astrocytoma, which differ in the presence of 1p/19q codeletion, were discerned with a correct rate of 81%.
Conclusion: These results demonstrate the feasibility of i) general brain tumor recognition and ii) extraction of diagnostic information by means of one analytical method. This opens the possibility to perform intraoperative optical biopsies using Raman spectroscopy for in situ delineation and diagnosis of brain tumors.
