SBIR Phase I: Automated Pattern Recognition in Images Produced by Comprehensive Two-Dimensional Gas Chromatography

这个小型企业创新研究第一阶段项目启动了对由全面的二维气相色谱产生的图像中的自动模式识别的严格研究。全二维气相色谱(GCxGC)是一种新兴的化学分离技术，其分离能力比传统GC成倍提高。随着性能的极大提高，GCxGC生成的数据量和复杂性都大大增加。GCxGC数据的数量和复杂性使得人类对GCxGC图像的分析既困难又耗时，并促使对自动化处理的需求。这一第一阶段项目进行了实验和理论研究，以使观察到的化学分离模式与人类专家注释的先前记录的模式相匹配的过程自动化。目标是为GCxGC模式匹配确定有前途的统计模型，并证明自动识别的可行性。在这项第一阶段的工作中，重要的预期成果包括GCxGC图像中模式变化和翘曲的统计特征，先前观察到的模式模板的有用注释的目录，以及自动模式识别的原型算法的开发。第一阶段的结果，表征了GCxGC模式，编目注释，并论证了自动化处理的可行性，将为第二阶段的研究提供基础，旨在开发商业标记的GCxGC方法。这项研究具有很高的潜在影响，各种应用。GC的商业应用包括石油、环境样品、食品和饮料、香料和毒素(如化学战剂)的分析。从GCxGC图像中自动识别化学成分的软件的提供将促进使用传统GC的实验室采用GCxGC技术，并将有助于开发需要卓越分离性能的新市场。

This Small Business Innovation Research Phase I project initiates rigorous investigation of automated pattern recognition in images produced by comprehensive two-dimensional gas chromatography. Comprehensive two-dimensional gas chromatography (GCxGC) is an emerging technology for chemical separation that provides a multiplicative increase in separation capacity over traditional GC. With this greatly increased performance, GCxGC generates data in significantly larger quantity and with significantly greater complexity. The quantity and complexity of GCxGC data makes human analyses of GCxGC images difficult and time-consuming and motivates the need for automated processing. This Phase I project undertakes both experimental and theoretical investigations into automating the process of matching observed patterns of chemical separations against previously recorded patterns annotated by human experts. The goals are to determine promising statistical models for GCxGC pattern matching and to demonstrate the feasibility of automated recognition. In this Phase I work, important anticipated results include statistical characterization of pattern variations and warping in GCxGC images, a catalog of useful annotations of previously observed pattern templates, and development of a prototype algorithm for automated pattern recognition. Phase I results, characterizing GCxGC patterns, cataloging annotations, and demonstrating the feasibility of automated processing, will provide a foundation for Phase II research aimed at developing commercially sign cant GCxGC methods.This research has high potential impact for a variety of applications. Commercial applications of GC include analyses of petroleum, environmental samples, foods and beverages, fragrances, and toxins (e.g., chemical warfare agents). The availability of software for automated recognition of chemical components from GCxGC images will facilitate adoption of GCxGC technology in laboratories using traditional GC and will contribute to the development of new markets, which require superior separation performance.