Multivariate optical computations (MOCs) offer improved analytical precision and increased speed of analysis via synchronous data collection and numerical computation with scanning spectroscopic systems. The improved precision originates in the redistribution of integration time from spurious channels to informative channels in an optimal manner for increasing the signal-to-noise ratio with multivariate analysis under the constraint of constant total analysis time. In this work, MOCs perform the multiplication and addition steps of spectral processing by adjusting the integration parameters of the optical detector or adjusting the scanning profile of the tunable optical filter. Improvement in the precision of analysis is achieved via the implicit optimization of the analytically useful signal-to-noise ratio. The speed improvements are realized through simpler data post-processing, which reduces the computation time required after data collection. Alternatively, the analysis time may be significantly truncated while still seeing an improvement in the precision of analysis, relative to competing methods. Surface plasmon resonance (SPR) spectroscopic sensors and visible reflectance spectroscopic imaging were used as test beds for assessing the performance of MOCs. MOCs were shown to reduce the standard deviation of prediction by 15% compared to digital data collection and analysis with the SPR and up to 45% for the imaging applications. Similarly, a 30% decrease in the total analysis time was realized while still seeing precision improvements.

No abstract

We describe the application of neural networks to a theoretical problem: the correction of inaccuracies in infrared spectra as predicted by the PM3 semiempirical method. Twenty-eight "peak-correcting" backpropagation neural networks were trained to predict the location of a characteristic infrared peak when given a scaled topological map of one of 1116 literature spectra. The infrared spectra of 200 aliphatics were then calculated using PM3, displayed graphically in Infrared Spectrum Comparison, and submitted to the appropriate "peak-correcting" neural network(s) based on a rule set implemented via an interface to HyperCube's HyperChem software. Results show an average 8-fold decrease in prediction error between PM3-predicted and neural network-corrected peak locations.