The application of high-resolution LC-MS experiments in drug discovery is very versatile due to high selectivity, sensitivity, and low sample volume requirements. A single HR LC-MS experiment is capable of quantitatively evaluating thousands of endogenous molecules as well as drug-derived metabolites. Advances in chromatographic separation, detector ruggedness and increased data acquisition rates enable the attainment of information-rich data sets which have shifted the bottleneck to data analysis.<br/>A primary concern of data analysis is the assessment of system suitability which reflects the quality of the collected signals. A common system suitability test evaluates the signals of characteristic peaks corresponding to a mixture of stable labeled internal standards spiked into a sample. These evaluations are driven by FDA guidelines and usually performed manually - an analyst examines the first and last injection as well as several randomly selected injections in between. A manual examination of each peak in a large study is not practical due to time and personnel constraints.<br/>Traditional statistics-based approaches to automation of system suitability testing require the calculation of averages of peak attributes such as area, symmetry, tailing factor, etc. Calculation of averaged peak attributes is sensitive to various factors including noise, system usage, and detector configuration and causes a lag in discovering abnormalities, finding them only when problems persist in a significant number of injections.<br/>This presentation discusses a novel approach to automation of data analysis which facilitates real-time signal processing during acquisition and utilizes a Convolutional Neural Network (CNN) model for peak image classification. Our approach uses images of characteristic peaks to detect abnormalities, bypassing the biases inherent in other methods of automatic monitoring. Automated data analysis algorithms gather the information relevant to mass spectroscopic characterization of molecules and report the quantitative composition of a biological sample while maintaining quality and time efficiency.<br/>

Real time monitoring of the suitability of high resolution LC-MS systems by image recognition algorithm using peaks of spiked-in internal standards

Macromolecular Chemistry: The Second Century

[Author Benefits](https://scimeetings.acs.org/?utm_source=pubs_content_marketing&utm_medium=website&utm_campaign=0320_MCF_NPI_Launch_Spring_Homepage&ref=pubs_content_marketing)

[How to use SciMeetings](https://storage.googleapis.com/dl.morressier.com/acs/SciMeetings-ACSSpring2020-HowToGuide_M3020-Final.pdf)

[Frequently Asked Questions](https://storage.googleapis.com/dl.morressier.com/acs/SciMtng_ACSNMSp2020_FAQ_M3120_Final.pdf)

ACS Spring 2020 National Meeting & Expo

The official archive and repository for all ACS Meetings & Expositions abstracts and select posters or presentations.