WAVELET-BASED AUTOMATED CHROMOSOME IDENTIFICATION

点击翻译按钮获取中文摘要

Commercial automated karyotyping instruments have improved to the point where the major factor limiting throughput is the time required for operator correction of chromosome classification errors. An improvement in chromosome classification accuracy would significantly increase the value of these instruments in cytogenetics labs. The goal of this project is to develop and commercialize significantly improved chromosome measurement and classification techniques for automated karyotyping. Currently the best-performing chromosome classification approach uses Weighted Density Distribution (WDD) features [11] to quantify the banding pattern of the chromosomes. These are computed as inner products between the banding profile and a set of WDD basis functions. The particular set of 1unctions originally proposed by Granum [11,38] has come into widespread use. In Phase I we showed that better function sets exist and that our new approach can find better WDD features than the best currently used. We have an innovative wavelet-based method for generating WDD functions and a chromosome classification testbed which supports large scale classification experiments. We propose to conduct a thorough, methodical search for better performing basis functions in Phase II. Phase III will incorporate the technology into PSI's PowerGene automated karyotyping instruments. PROPOSED COMMERCIAL APPLICATIONS: When the new chromosome classification technology is qualified for routine application, it will be incorporated into PSI's Powergene products, both in new systems sold and as an upgrade to existing systems.