Subspace modeling plays an important role in face recognition. Independent Component Analysis （ICA）, a multivariable statistical analysis technique, can be seen as an extension of traditional Principal Com- ponent A...Subspace modeling plays an important role in face recognition. Independent Component Analysis （ICA）, a multivariable statistical analysis technique, can be seen as an extension of traditional Principal Com- ponent Analysis （PCA） technique, which addresses high order statistics as well as second order statistics. In this paper, a new scheme of subspace-based representation called Discriminant Independent Component Analysis （DICA） is proposed, which combines the strength＂ of unsupervised learning of ICA and supcrvised learning of Linear Discriminant Analysis （LDA）, and efficiently enhances the generalization ability of ICA-based representation method. Based on DICA subspace analysis, a set of optimal vectors called ＂discriminant independent faces＂ are learned from face samples. The effectiveness of our method is demonstrated by performance comparisons with some popular methods such as ICA, PCA, and PCA＋LDA. On the large scale database of IIS, significant improvements are observed when there are fewer training samples per person available.展开更多
The moving window bis corelation coefficients(MW BiCC)was proposed and employed for the discriminant analysis of transgenic sugarcane leaves and B-thalassemia with visible and near-infrared(Vis NIR)spectroscopy.The we...The moving window bis corelation coefficients(MW BiCC)was proposed and employed for the discriminant analysis of transgenic sugarcane leaves and B-thalassemia with visible and near-infrared(Vis NIR)spectroscopy.The well-performed moving window principal component analysis linear discriminant analysis(MWPCA-LDA)was also conducted for comparison.A total of 306 transgenic(positive)and 150 nont ransgenic(negative)leave samples of sugarcane were collected and divided to calibration,prediction,and validation.The diffuse reflection spectra were corected using Savitzky-Golay(SG)smoothing with first-order derivative(d=1),third-degree polynomial(p=3)and 25 smpothing points(m=25).The selected waveband was 736-1054nm with MW-BiCC,and the positive and negative validation recognition rates(V_REC^(+),VREC^(-))were 100%,98.0%,which achieved the same effect as MWPCA-LDA.Another example,the 93 B-thalassemia(positive)and 148 nonthalassemia(negative)of human hemolytic samples were colloctod.The transmission spectra were corrected using SG smoothing withd=1,p=3 and m=53.Using M W-BiCC,many best wavebands were selected(e.g.,1116-1146,17941848 and 22842342nm).The V_REC^(+)and V_REC^(-)were both 100%,which achieved the same effect as MW-PCA-LDA.Importantly,the BICC only required ca lculating correlation cofficients between the spectrum of prediction sample and the average spectra of two types of calibration samples.Thus,BiCC was very simple in algorithm,and expected to obtain more applications.The results first confirmed the feasibility of distinguishing B-thalassemia and normal control samples by NIR spectroscopy,and provided a promising simple tool for large population thalassemia screening.展开更多
Concurrent chemoradiotherapy(CCRT)is the choice of treatment for locally adv anced cervical cancers;however,tumors exhibit diverse response to treatment.Early prediction of tumor response leads to individualizing trea...Concurrent chemoradiotherapy(CCRT)is the choice of treatment for locally adv anced cervical cancers;however,tumors exhibit diverse response to treatment.Early prediction of tumor response leads to individualizing treatment regimen.Response evaluation criteria in solid tumors(RECIST),the current modality of tumor response assessment,is often subject ive and carried out at the first visit after treatment,which is about four months.Hence,there is a need for better predictive tool for radioresponse.Optical spectroscopic techniques,sensitive to molecular alteration,are being pursued as potential diagnostic tools.Present pilot study ains to explore the fiber-optic-based Raman spectroscopy approach in prediction of tumor response to CCRT,before taking up extensive in vivo studies.Ex vivo Raman spectra were acquired from biopsies collected from 11 normal(148 spectra),16 tumor(201 spectra)and 13 complete response(151 CR spectra),one partial response(8 PR spectra)and one nonresponder(8 NR spectra)subjects.Data was analyzed using principal component linear discriminant analysis(PC-LDA)followed by leave-one-out cross-validation(LOO-CV).Findings suggest that normal tissues can be efficiently classified from both pre-and post-treated tumor biopsies,while there is an overlap between pre-and post-CCRT tumor tissues.Spectra of CR,PR and NR tissues were subjected to principal component analysis(PCA)and a tendency of classification was observed,corroborating previous studies.Thus,this study further supports the feasibility of Raman spectroscopy in prediction of tumor radioresponse and prospective noninvasive in vivo applications.展开更多
基金Supported by the Key Project of the National Natural Science Foundation of China(No.90104030)the National Natural Science Foundation of China(No.60401015)
文摘Subspace modeling plays an important role in face recognition. Independent Component Analysis （ICA）, a multivariable statistical analysis technique, can be seen as an extension of traditional Principal Com- ponent Analysis （PCA） technique, which addresses high order statistics as well as second order statistics. In this paper, a new scheme of subspace-based representation called Discriminant Independent Component Analysis （DICA） is proposed, which combines the strength＂ of unsupervised learning of ICA and supcrvised learning of Linear Discriminant Analysis （LDA）, and efficiently enhances the generalization ability of ICA-based representation method. Based on DICA subspace analysis, a set of optimal vectors called ＂discriminant independent faces＂ are learned from face samples. The effectiveness of our method is demonstrated by performance comparisons with some popular methods such as ICA, PCA, and PCA＋LDA. On the large scale database of IIS, significant improvements are observed when there are fewer training samples per person available.
基金supported by the Science and Technology Project of Guangdong Province of China(Nos.2014A020213016 and 2014A020212445).
文摘The moving window bis corelation coefficients(MW BiCC)was proposed and employed for the discriminant analysis of transgenic sugarcane leaves and B-thalassemia with visible and near-infrared(Vis NIR)spectroscopy.The well-performed moving window principal component analysis linear discriminant analysis(MWPCA-LDA)was also conducted for comparison.A total of 306 transgenic(positive)and 150 nont ransgenic(negative)leave samples of sugarcane were collected and divided to calibration,prediction,and validation.The diffuse reflection spectra were corected using Savitzky-Golay(SG)smoothing with first-order derivative(d=1),third-degree polynomial(p=3)and 25 smpothing points(m=25).The selected waveband was 736-1054nm with MW-BiCC,and the positive and negative validation recognition rates(V_REC^(+),VREC^(-))were 100%,98.0%,which achieved the same effect as MWPCA-LDA.Another example,the 93 B-thalassemia(positive)and 148 nonthalassemia(negative)of human hemolytic samples were colloctod.The transmission spectra were corrected using SG smoothing withd=1,p=3 and m=53.Using M W-BiCC,many best wavebands were selected(e.g.,1116-1146,17941848 and 22842342nm).The V_REC^(+)and V_REC^(-)were both 100%,which achieved the same effect as MW-PCA-LDA.Importantly,the BICC only required ca lculating correlation cofficients between the spectrum of prediction sample and the average spectra of two types of calibration samples.Thus,BiCC was very simple in algorithm,and expected to obtain more applications.The results first confirmed the feasibility of distinguishing B-thalassemia and normal control samples by NIR spectroscopy,and provided a promising simple tool for large population thalassemia screening.
文摘Concurrent chemoradiotherapy(CCRT)is the choice of treatment for locally adv anced cervical cancers;however,tumors exhibit diverse response to treatment.Early prediction of tumor response leads to individualizing treatment regimen.Response evaluation criteria in solid tumors(RECIST),the current modality of tumor response assessment,is often subject ive and carried out at the first visit after treatment,which is about four months.Hence,there is a need for better predictive tool for radioresponse.Optical spectroscopic techniques,sensitive to molecular alteration,are being pursued as potential diagnostic tools.Present pilot study ains to explore the fiber-optic-based Raman spectroscopy approach in prediction of tumor response to CCRT,before taking up extensive in vivo studies.Ex vivo Raman spectra were acquired from biopsies collected from 11 normal(148 spectra),16 tumor(201 spectra)and 13 complete response(151 CR spectra),one partial response(8 PR spectra)and one nonresponder(8 NR spectra)subjects.Data was analyzed using principal component linear discriminant analysis(PC-LDA)followed by leave-one-out cross-validation(LOO-CV).Findings suggest that normal tissues can be efficiently classified from both pre-and post-treated tumor biopsies,while there is an overlap between pre-and post-CCRT tumor tissues.Spectra of CR,PR and NR tissues were subjected to principal component analysis(PCA)and a tendency of classification was observed,corroborating previous studies.Thus,this study further supports the feasibility of Raman spectroscopy in prediction of tumor radioresponse and prospective noninvasive in vivo applications.
