TY - GEN
T1 - NNLSF
T2 - 12th IEEE International Symposium on Biomedical Imaging, ISBI 2015
AU - Yao, Shun
AU - Chang, Cheng
AU - Xu, Wei
AU - Zhou, Naiyun
AU - Chen-Wiegart, Yu Chen Karen
AU - Wang, Jiajun
AU - Wang, Jun
AU - Yu, Dantong
N1 - Publisher Copyright:
© 2015 IEEE.
PY - 2015/7/21
Y1 - 2015/7/21
N2 - Full field X-ray spectroscopy imaging in NSLS-II will provide unprecedented insights into 2D/3D chemical compositions of nanomaterials. Spectra fitting which decomposes the experimental spectra data into chemical compositions plays a key role in the technique. Existing fitting methods including Brute Force (BF) and Constrained Least Square Fitting (CLSF) rest upon fitting fractions and suffer two problems: 1) loss of the thickness information; 2) demands for filtering. In this paper, we propose a new fitting method Non-Negative Least Square Fitting (NNLSF), which directly fit thicknesses instead of fractions. Our experiments in both simulation and real datasets show that, NNLSF 1) provides more information in fitting results than current approaches, 2) saves the efforts of filtering, and also 3) is 6∼96 times faster than alternatives. All the methods (BF, CLSF, NNLSF) were implemented as open-source software with a friendly GUI.
AB - Full field X-ray spectroscopy imaging in NSLS-II will provide unprecedented insights into 2D/3D chemical compositions of nanomaterials. Spectra fitting which decomposes the experimental spectra data into chemical compositions plays a key role in the technique. Existing fitting methods including Brute Force (BF) and Constrained Least Square Fitting (CLSF) rest upon fitting fractions and suffer two problems: 1) loss of the thickness information; 2) demands for filtering. In this paper, we propose a new fitting method Non-Negative Least Square Fitting (NNLSF), which directly fit thicknesses instead of fractions. Our experiments in both simulation and real datasets show that, NNLSF 1) provides more information in fitting results than current approaches, 2) saves the efforts of filtering, and also 3) is 6∼96 times faster than alternatives. All the methods (BF, CLSF, NNLSF) were implemented as open-source software with a friendly GUI.
KW - chemical mapping
KW - non-negative least square fitting
KW - NSLS-II
KW - XANES
UR - https://www.scopus.com/pages/publications/84944322069
U2 - 10.1109/ISBI.2015.7164077
DO - 10.1109/ISBI.2015.7164077
M3 - Conference contribution
AN - SCOPUS:84944322069
T3 - Proceedings - International Symposium on Biomedical Imaging
SP - 1155
EP - 1158
BT - 2015 IEEE 12th International Symposium on Biomedical Imaging, ISBI 2015
PB - IEEE Computer Society
Y2 - 16 April 2015 through 19 April 2015
ER -