TY - GEN
T1 - De-noising and event extraction for silicon pore sensors using matrix decomposition
AU - Sattigeri, P.
AU - Thiagarajan, J. J.
AU - Natesan Ramamurthy, K.
AU - Spanias, Andreas
AU - Goryll, Michael
AU - Thornton, Trevor
PY - 2012
Y1 - 2012
N2 - Silicon pores with diameters in the range of micro/nano-meters can be used to detect an array of analytes. Silica beads are used as carriers of biomolecules through the pores. Passage of beads through the pores are termed as translocation events. In the presence of certain pairs of biomolecules, the pores exhibit trapping behaviour where the pores gets partially blocked. Such behaviour is termed as a trapping event. In this paper, we analyze simulated data of silicon-pore sensors and propose methods to perform signal de-noising and extraction of translocation/trapping events. In the first approach, we use the Discrete Wavelet based de-noising (DWT) as a preprocessing step. We window the signal and stack the segment into a matrix. The data matrix is decomposed into low rank and non-positive sparse components using the modified RPCA (Robust Principal Component Analysis) algorithm. In the second approach, we decompose the noisy signal matrix obtained without DWT. A GoDec (Go Decomposition) based approach is used here, with an explicit noise component and additionally a smoothness constraint. We compare both approaches and show results for signal de-noising and translocation/trapping event extraction.
AB - Silicon pores with diameters in the range of micro/nano-meters can be used to detect an array of analytes. Silica beads are used as carriers of biomolecules through the pores. Passage of beads through the pores are termed as translocation events. In the presence of certain pairs of biomolecules, the pores exhibit trapping behaviour where the pores gets partially blocked. Such behaviour is termed as a trapping event. In this paper, we analyze simulated data of silicon-pore sensors and propose methods to perform signal de-noising and extraction of translocation/trapping events. In the first approach, we use the Discrete Wavelet based de-noising (DWT) as a preprocessing step. We window the signal and stack the segment into a matrix. The data matrix is decomposed into low rank and non-positive sparse components using the modified RPCA (Robust Principal Component Analysis) algorithm. In the second approach, we decompose the noisy signal matrix obtained without DWT. A GoDec (Go Decomposition) based approach is used here, with an explicit noise component and additionally a smoothness constraint. We compare both approaches and show results for signal de-noising and translocation/trapping event extraction.
KW - Analyte classification
KW - De-noising
KW - Matrix decomposition
KW - Silicon pore sensors
UR - http://www.scopus.com/inward/record.url?scp=84880153477&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=84880153477&partnerID=8YFLogxK
U2 - 10.1049/ic.2012.0098
DO - 10.1049/ic.2012.0098
M3 - Conference contribution
AN - SCOPUS:84880153477
SN - 9781849197120
T3 - IET Seminar Digest
BT - Sensor Signal Processing for Defence, SSPD 2012
T2 - Sensor Signal Processing for Defence, SSPD 2012
Y2 - 25 September 2012 through 27 September 2012
ER -