Performance assessment of probe-based Raman spectroscopy systems for biomedical analysis

We present a methodology for evaluating the performance of probe-based Raman spectroscopy systems for biomedical analysis. This procedure uses a biological standard sample and data analysis approach to circumvent many of the issues related to accurately measuring and comparing the signal quality of Raman spectra between systems. Dairy milk is selected as the biological standard due to its similarity to tissue spectral properties and because its homogeneity eliminates the dependence of probe orientation on the measured spectrum. A spectral dataset is first collected from milk for each system configuration, followed by a model-based correction step to remove photobleaching artifacts and accurately calculate SNR. Results demonstrate that the proposed strategy, unlike current methods, produces an experimental SNR that agrees with the theoretical value. Four preconfigured imaging spectrographs that share similar manufacturer specifications were compared, showing that their capabilities to detect biological Raman spectra widely differ in terms of throughput and stray light rejection. While the methodology is used to compare spectrographs in this case, it can be adapted for other purposes, such as optimizing the design of a custom-built Raman spectrometer, evaluating inter-probe variability, or examining how altering system subcomponents affects signal quality.