Machine learning for optical chemical multi-analyte imaging

Abstract

Simultaneous sensing of metabolic analytes such as pH and O2 is critical in complex and heterogeneous biological environments where analytes often are interrelated. However, measuring all target analytes at the same time and position is often challenging. A major challenge preventing further progress occurs when sensor signals cannot be directly correlated to analyte concentrations due to additional effects, overshadowing and complicating the actual correlations. In fields related to optical sensing, machine learning has already shown its potential to overcome these challenges by solving nested and multidimensional correlations. Hence, we want to apply machine learning models to fluorescence-based optical chemical sensors to facilitate simultaneous imaging of multiple analytes in 2D. We present a proof-of-concept approach for simultaneous imaging of pH and dissolved O2 using an optical chemical sensor, a hyperspectral camera for image acquisition, and a multi-layered machine learning model based on a decision tree algorithm (XGBoost) for data analysis. Our model predicts dissolved O2 and pH with a mean absolute error of < 4.50·10−2 and < 1.96·10−1, respectively, and a root mean square error of < 2.12·10−1 and < 4.42·10−1, respectively. Besides the model-building process, we discuss the potentials of machine learning for optical chemical sensing, especially regarding multi-analyte imaging, and highlight risks of bias that can arise in machine learning-based data analysis.

Multiparameter Sensing of Oxygen and pH at Biological Interfaces via Hyperspectral Imaging of Luminescent Sensor Nanoparticles

“…Despite good spectral separation between the emission of sensor and references dyes and sample fluorescence, we still found that minor overlaps between the sensor emission and Chl fluorescence and/or sample scattering could affect the calibration and sensor performance, therefore calling for sample-specific calibration procedures. Zieger and Koren 21 demonstrated how machine-learning algorithms can be employed to optimize the precision of combined imaging of O 2 and pH in simple calibration experiments. While such an approach was beyond the scope of the present paper, machine learning could potentially also be used to help optimize sensor calibration in more complex settings, where sample fluorescence and scattering affect the ratiometric read-out.…”

“…Subsequent deconvolution of the hyperspectral data enabled O 2 imaging at a high dynamic range with the combined sensor material. Zieger and Koren 21 also demonstrated the use of machine-learning algorithms for optimizing simultaneous hyperspectral imaging of pH and O 2 based on the analysis of detailed calibration experiments.…”

Optical sensors (optodes) for multiparameter chemical imaging: classification, challenges, and prospects

“…In this context, we have already shown that depending on the strength of the optical overlap of the different receptors, signal deconvolution can be achieved either with simpler algorithms such as spectral linear combination and least-square fitting 72 or with more complex machine learning algorithms. 74…”

“…Machine learning approaches offer thereby a great opportunity to process these data sets and translate them into valuable, multiparametric information. 74,105–107 Moreover, in recent years, more and more deep learning algorithms have been reported not only for image classification but also for multiparameter regression purposes. 108 For more details on recent advances in chemometric calibration methods in modern spectroscopy, the interested reader is referred to the respective review by Wang et al 79…”