Meta-Analysis of the Efficacy of Raman Spectroscopy and Machine-Learning-Based Identification of Glioma Tissue

Area of Science:

• Neuroscience
• Medical Technology
• Oncology

Background:

• Intraoperative Raman spectroscopy (RS) offers rapid, noninvasive tissue differentiation.
• Previous meta-analyses were limited by older technology and methods.
• Advancements in spectroscopy and machine learning may enhance RS efficacy.

Purpose of the Study:

• To conduct a meta-analysis assessing the efficacy of intraoperative RS for differentiating glioma from normal brain tissue.
• To update previous findings with recent studies (2016-2022).

Main Methods:

• Systematic literature search on PubMed and Web of Science for relevant prospective and retrospective studies.
• Inclusion of studies using intraoperative RS and standard histology for tissue differentiation.
• Meta-analysis of diagnostic odds ratios, sensitivity, and specificity using a random-effects model in JASP.

Main Results:

• Nine studies comprising 673 patients and 8319 spectra were analyzed.
• A diagnostic odds ratio of 76.71 indicated high efficacy, despite significant heterogeneity.
• Overall sensitivity was 95.3% and specificity was 71.2%, with an AUC of 0.9265.

Conclusions:

• Intraoperative Raman spectroscopy is a highly promising tool for surgeons.
• RS enables quick and accurate differentiation between glioma and healthy brain tissue.
• The findings support the integration of RS into surgical practice for improved glioma resection.