Automated image curvature assessment and correction for high-throughput Raman spectroscopy and microscopy

Abstract

Raman spectroscopy and microscopy can provide molecular information for complex materials such as biological tissue and cells. In these applications, light-collection throughput is essential for speedy acquisition of high-quality data. To im- prove throughput, two-dimensional detectors and high numerical aperture (NA) optical systems have been employed. However, owing to the out-of-plane diffraction in grating-based dispersive spectrograph, the entrance slit image formed at the detector plane is curved along the vertical direction. Direct vertical binning of individual detector rows without correcting the curvature results in degraded spectral resolution and peak misalignment. We evaluate two software approaches to remove the image cur- vature after high-throughput data acquisition, with the objective to retain instrument spectral resolution and peak accuracy as if a linear-array detector were used. Curvature correction and detection are achieved in two steps: calibration of the image curva- ture using a Raman active material and application of the correction to future curved images. This method has been employed for a high-NA, large CCD Raman spectroscopic system deigned for non-invasive glucose sensing, a medium-NA, medium-size CCD line-scan Raman microscope designed for high-throughput tissue and cellular imaging, and an active-illumination Raman microscope. We show that remarkable improvement in data fidelity can be obtained as assessed by peak misalignment, distri- bution of data variance, and the waveform of principal component spectra. High quality curvature correction is essential for quantitative analysis such as the multivariate calibration, spectral pattern recognition, and peak shift detection based techniques. The software approach is highly flexible for instrument modification.