Cancellation of Spectral and Spatial Crosstalk in Spectral Imaging for High-Dynamic-Range Electrophoretic Analysis of...

In spectral imaging, an optical system generates two mutually dependent kinds of crosstalk on an image sensor: spectral crosstalk (i.e., spectral mixing) between fluorescences of different dyes and spatial crosstalk (i.e., image artifacts) between fluorescences from different emission points in the field of view of the sensor. Therefore, an algorithm to cancel both kinds of crosstalk simultaneously (i.e., simultaneous spectral unmixing and image-artifact reduction) is proposed. The algorithm is based on the assumption that a crosstalk matrix (i.e., point-spread functions, PSFs) consisting of all crosstalk ratios is constant regardless of the causes of both kinds of crosstalk. By applying the algorithm to a nine-wavelength-band measurement of four-capillary electrophoretic separation of STR-PCR (short tandem repeat-polymerase chain reaction) products labeled with six dyes, true peaks of each of the dyes were obtained, while false peaks due to spatial crosstalk were reduced below the lower limit of detection in electropherograms. As a result, effective sensitivity and effective dynamic range were improved by 2 orders of magnitude. Moreover, it became possible to perform robust human identification of on-site-collected trace samples containing template DNA at any concentration in a 3-order concentration range by a single STR-PCR and a single electrophoretic separation.