Contribution of the LUT University to the infrastructure includes development and establishment of an open-access quantitative spectral data processing and image generation platform. The user-friendly data-analysis tools are available on an open-access online platform webpage developed for the infrastructure.

The platform is essential for optimising research outputs with the imaging equipment. It addresses four key requirements: a) big data handling, b) quantitative coherent Raman and IR spectral data analysis, c) user-friendly and flexible image generation, and d) remote access. The data analysis algorithms have been tailored to coherent Raman (CARS and SRS) and near-field IR spectral data. Crucially, quantitative imaging with CARS data has been largely prevented by the nonlinear and coherent nature of the CARS signal. The LUT team has overcome this barrier by pioneering specific wavelet and numerical phase retrieval approaches to extract the linear vibrational contribution in the multiplex CARS signal (Raman line-shape). Quantitative analysis of these processed CARS, SRS and near-field IR data have been optimized and developed with supervised and unsupervised methods including machine learning, such that they can be readily applied and used to generate chemical and structural images by end-users.

The main contributions for the online data analysis platform and computational tool implementations include the following:

• Frontend for HARKANA analysis platform
• Sequential Monte Carlo for Line-shape Optimized Maximum Entropy line narrowing
• Log-Gaussian gamma process parameter estimation for corrected CARS and Raman measurements
• Partially Bayesian neural network with a gamma-distributed output layer for correcting spectroscopic measurements
• Interpolated inverse discrete wavelet transforms for spectral background corrections
• qCSI data analysis platform template