Applications

Off-the-Shelf Applications

Our software products have been used in a wide range of applications. The following list gives just a few examples. Our staff has experience in many of these fields and can provide world-class consulting for your project.

Detection of out-of-specification product
Identification of end-point in batch reactions
Close-loop process control of continuous reactions
Monitoring of pharmaceutical drying processes
Identification/Classification of incoming materials
Detection of semiconductor failure
Polymer compositional analyses (copolymer, blend, emulsions, additives, endgroups, etc)
Pharmaceutical solid state analyses (polymorph identification, blend uniformity, solid dosage form, etc)
Pure component spectra, identification of intermediates, and chemical process understanding of reaction spectra
Mitigate or eliminate the potentially deleterious effects of interferences
Allows the use of non-specific detectors (e.g., NIR)
Identifying and quantifying dissolved organic matter compounds
Resolving overlapping peaks in LC and GC
Determining fat content non-invasively in minced meat using low-resolution NMR
Quantifying dioxin in fish products
Characterizing oxidation products in cheese and butter
Real-time monitoring and chemical profiling of a cultivation process
Determining regional differences in world human body dimensions
Predicting toxicity from sediment chemistry
Evaluation of pharmacological responses in early clinical trials

Custom Applications

Eigenvector Research’s staff has decades of experience in developing real solutions to complicated problems. Many of our consulting applications involve providing some sort of software end-product and we make use of our staff’s extensive experience to offer high-quality, custom solutions. These solutions range from simple scripts to sophisticated custom GUIs. Below is a short description of some example projects we’ve worked on. Click on any item to see a description of the solution we offered. If you have a question about how Eigenvector Research’s consulting staff can help you, contact us here.

Model export scheme for embedded application.

Model_Exporter was used to convert multivariate models generated by PLS_Toolbox into self-contained, secure TCL-coded models that are deployable on a low-overhead PC board. In this role, Model_Exporter was utilized as an integral part of a streamlined model development and maintenance process, to support a network of distributed, remote analyzers.

High-throughput real-time prediction engine.

An autonomous prediction engine was developed to accept connections from existing in-house data management system. The engine has a simple-to-use configuration GUI which accepts standard PLS_Toolbox/Solo models and applies these models in real-time returning the results to the data management system for display and storage.

Custom web-based prediction engine.

A web-server was developed for plant personnel to view model predictions in real-time through a highly interactive and customizable web page. Based on our Solo_Predictor engine and AJAX technology, the web page automatically queries a central database and applies specified PLS_Toolbox/Solo models to the data. The results are displayed in tabular or graphical format with the ability to “drill-down” to contributions and/or raw data. The interface includes high-level “Manager” views, mid-level “Engineer” views, and expert-level “Detail” views.

Customized Solo for a pharmaceutical laboratory application.

Client-supplied clustering and visualization algorithms were added to our standard Solo package along with application-specific flowcharts (to help less-experienced lab personnel run the system). The package is intentionally limited in the methods which are available so as to help guide users to the specific actions necessary for the application. The result is a package which the client could easily distribute within their research division with minimal training requirements.

The custom application makes use of our flexible flowcharting frame among many other customizable features of our GUIs.

Customized Solo for bundling with semiconductor instrumentation.

A highly customized version of Solo was supplied for bundling with a semiconductor manufacturing company’s instruments. This package included a custom automatic data importing routine, specialized preprocessing options, application-specific automatic analysis tools developed by Eigenvector Research, command-line utility to automatically generate models, and custom import/export format for models.

Biological system automated preprocessing and modeling scripts.

Working closely with the client, Eigenvector Research developed application-specific preprocessing scripts to help automate model development and evaluation for a complicated biological system. Eigenvector’s experience in real-life applications, scientific instrumentation, experimental design, and chemical and biological systems allowed us to work in concert with the client’s technical staff. These scripts were developed using PLS_Toolbox for Matlab and included training of the client’s staff on their use as well as development of calibration protocols.

Novel library search algorithm for embedded application.

A custom clustering and classification search algorithm was developed for a hand-held device. The algorithms were specifically developed in close contact with client engineers to allow implementation on a low-memory/low-power embedded computer and easy expansion to additional applications.

Customized high-speed model deployment.

Model deployment software was designed for, and integrated with, an existing customized data-logging client to support a sensor array development project. This custom software, which utilized Matlab engine functions, enabled parallel deployment and evaluation of multiple calibration models at a high execution rate (> 1 Hz).

Remote-sensing hyperspectral scene generator.

EVRI developed a hyperspectral scene generator for testing of remote sensing algorithms. The “InfraRed Systems in General Environments (IR-SAGE)” platform was created to incorporate the real-life physics of remote-sensing and library spectra. Our staff developed the mathematical models and graphical environment to create sophisticated test data which greatly reduced the development time for hyperspectral models.

D. M. Sheen, N. B. Gallagher, P. G. Heasler, J. F. Schultz, B. M. Wise, S. W. Sharpe, and K. K. Anderson, “Infrared Chemical Detection Systems Modeling and Advanced Chemometric Analysis”, PNNL 13737, Pacific Northwest National Laboratory, Richland (2001).

Sheen, D.M., Gallagher, N.B., Sharpe, S.W., Anderson, K.K., and Shultz, J.F., “Impact of background and atmospheric variability on infrared hyperspectral chemical detection sensitivity”, SPIE Proceedings, 5093, 218-229 (2003).

Gallagher, N.B., Sheen, D.M., Shaver, J.M., Wise, B.M. and Shultz, J.F., “Estimation of trace vapor concentration-pathlength in plumes for remote sensing applications from hyperspectral images”, SPIE Proceedings, 5093, 184-194 (2003).

N. B. Gallagher, B. M. Wise, and D. M. Sheen, “Error Analysis for Estimation of Trace Vapor Concentration-Pathlength in Stack Plumes”, Appl. Spec., 57(6), (2003).

Gallagher, N.B., Wise, B.M., and Sheen. D.M., “Estimation of Trace Vapor Concentration-Pathlength in Plumes for Remote Sensing Applications from Hyperspectral Images,” Anal. Chim. Acta., 490, 139-152 (2003).

Batch and semiconductor end-point detection.

A patented end-point detection system was developed for batch and semi-conductor applications.

Harvey, K.C., Hosch, J.W., Gallagher, N.B., and Wise, B.M., “System and method for determining endpoint in etch processes using partial least squares discriminant analysis in the time domain of optical emission spectra”, US Patent No. 6,830,939; Dec. 14 (2004).

Target testing algorithm for biological mixtures.

A novel method was developed to determine the components in complex biological mixtures using target testing. The method determines a list of possible components and then optimizes and eliminates similar compounds through exhaustive testing of all possible mixtures. The final algorithm was patented and written-up for use in a commercial application.

R. Schweitzer, P. J. Treado, W. Windig, Method for identifying components of a mixture via spectral analysis, US Patent 7,072,770, July 4, 2006

Controlling manufacturing in a validated environment.

By using a combination of multivariate and univariate laboratory measurements and providing a targeted output, this custom GUI interface calculates the operating parameters to set for this step in the process. The user interface was designed for use in a validated environment around strict specifications for record integrity.

Custom GUI for predicting solid-state impurities.

This application predicts the level of an impurity in a solid state material along with relevant statistical figures of merit. The interface generates a prediction model using an automatic rules-based procedure and then allows application of this model to future measurements. Traceability of the model lineage is transparent throughout, and the application ascertains users’ privileges to limit access to key functional capabilities.

Real-time monitoring and chemical profiling of a cultivation process.

A method for at-line quality assessment of a cultivation process was developed to (1) enable improved process control, (2) enable faster detection of batch end point, and (3) enable immediate quality assessment of final product. Fluorescence excitation-emission measurements were used and modeled by a PARAFAC model, providing a chemically interpretable visualization of the process variation.

Paper: http://dx.doi.org/10.1016/j.chemolab.2006.04.022