Parallel Factor Model Research Articles

Impact of dataset diversity on accuracy and sensitivity of parallel factor analysis model of dissolved organic matter fluorescence excitation-emission matrix.

Parallel factor (PARAFAC) analysis enables a quantitative analysis of excitation-emission matrix (EEM). The impact of a spectral variability stemmed from a diverse dataset on the representativeness of the PARAFAC model needs to be examined. In this study, samples from a river, effluent of a wastewater treatment plant, and algae secretion were collected and subjected to PARAFAC analysis. PARAFAC models of global dataset and individual datasets were compared. It was found that the peak shift derived from source diversity undermined the accuracy of the global model. The results imply that building a universal PARAFAC model that can be widely available for fitting new EEMs would be quite difficult, but fitting EEMs to existing PARAFAC model that belong to a similar environment would be more realistic. The accuracy of online monitoring strategy that monitors the fluorescence intensities at the peaks of PARAFAC components was examined by correlating the EEM data with the maximum fluorescence (Fmax) modeled by PARAFAC. For the individual datasets, remarkable correlations were obtained around the peak positions. However, an analysis of cocktail datasets implies that the involvement of foreign components that are spectrally similar to local components would undermine the online monitoring strategy.

Climate-driven terrestrial inputs in ultraoligotrophic mountain streams of Andean Patagonia revealed through chromophoric and fluorescent dissolved organic matter

Fluvial networks transport a substantial fraction of the terrestrial production, contributing to the global carbon cycle and being shaped by hydrologic, natural and anthropogenic factors. In this investigation, four Andean Patagonian oligotrophic streams connecting a forested catchment (~125km2) and draining to a double-basin large and deep lake (Lake Moreno complex, Northwestern Patagonia), were surveyed to analyze the dynamics of the allochthonous subsidy. The results of a 30month survey showed that the catchment supplies nutrients and dissolved organic matter (DOM) to the streams. The eruption of the Puyehue–Cordón Caulle at the beginning of the study overlapped with seasonal precipitation events. The largest terrestrial input was timed with precipitation which increased particulate materials, nutrients and DOM through enhanced runoff. Baseline suspended solids and nutrients were very low in all the streams (suspended solids: ~1mg/L; total nitrogen: ~0.02mg/L; total phosphorus: ~5μg/L), increasing several fold with runoff. Baseline dissolved organic carbon concentrations (DOC) ranged between 0.15 and 1mg/L peaking up to three-fold. Chromophoric and fluorescent analyses characterized the DOM as of large molecular weight and high aromaticity. Parallel factor modeling (PARAFAC) of DOM fluorescence matrices revealed three components of terrestrial origin, with certain degree of microbial processing: C1 and C2 (terrestrial humic-like compounds) and C3 (protein-like and pigment derived compounds). Seasonal changes in MOD quality represent different breakdown stages of the allochthonous DOM. Our survey allowed us to record and discuss the effects of the Puyehue–Cordón Caulle eruption, showing that due to the high slopes, high current and discharge of the streams the volcanic material was rapidly exported to the Moreno Lake complex. Overall, this survey underscores the magnitude and timing of the allochthonous input revealing the terrestrial subsidy to food webs in Patagonian freshwaters, which are among the most oligotrophic systems of the world.

Semi‐blind parallel factor based receiver for joint symbol and channel estimation in amplify‐and‐forward multiple‐input multiple‐output relay systems

In this study, the authors consider a new space-time coding scheme for two-hop amplify-and-forward (AF) relay multiple-input multiple-output (MIMO) relay communication systems. This scheme, called multiple Khatri–Rao space-time coding, which combines symbol stream blocking, spatial precoding and time-domain spreading. The authors show that the received signals at the destination node can be formulated as two parallel factor (PARAFAC) models, and then a novel semi-blind receiver is derived using a two-stage iterative fitting algorithm for joint symbol and channel estimation. Under the assumption that channel state information (CSI) is not available neither at the relay nodes nor at the destination node, the proposed semi-blind receiver operates close to the zero forcing receiver with perfect CSI. Moreover, without any channel training sequence, the proposed semi-blind receiver provides the destination node with full knowledge of all channel matrices involved in the transmission. Simulation results illustrate the performance of the proposed semi-blind receiver in comparison with alternative approaches.

2D-DOA and Mutual Coupling Estimation in Vehicle Communication System via Conformal Array

Many direction-of-arrival (DOA) estimation algorithms have been proposed recently. However, the effect of mutual coupling among antenna elements has not been taken into consideration. In this paper, a novel DOA and mutual coupling coefficient estimation algorithm is proposed in intelligent transportation systems (ITS) via conformal array. By constructing the spectial mutual coupling matrix (MCM), the effect of mutual coupling can be eliminated via instrumental element method. Then the DOA of incident signals can be estimated based on parallel factor (PARAFAC) theory. The PARAFAC model is constructed in cumulant domain using covariance matrices. The mutual coupling coefficients are estimated based on the former DOA estimation and the matrix transformation between MCM and the steering vector. Finally, due to the drawback of the parameter pairing method in Wan et al., 2014, a novel method is given to improve the performance of parameter pairing. The computer simulation verifies the effectiveness of the proposed algorithm.

Novel augmented parallel factor model for four-way calibration of high-performance liquid chromatography–fluorescence excitation–emission data

A new augmented parallel factor analysis model (Augmented PARAFAC) is presented, inspired by the useful augmentation concept employed in multivariate curve resolution-alternating least-squares (MCR-ALS), applicable to calibration based on non-quadrilinear four-way data, such as those produced by high-performance liquid chromatography with matrix excitation–emission fluorescence detection. The new model involves creating an augmented three-way array in the elution time direction, containing data for the calibration sample set and for each of the test samples, subsequently analyzed with an Augmented PARAFAC version. To test the properties of this approach, chromatographic data were simulated with different degrees of overlapping and misalignment among the chromatographic peaks. Additionally, experimental data from olive oil samples were tested with the new model, aimed at the quantitation of the level of chlorophylls and pheophytins. The results were compared with those obtained by data processing with MCR-ALS. Relative prediction errors (%) were: Augmented PARAFAC, 9.7, 21.0, 14.7 and 9.3, and MCR-ALS, 5.9, 14.5, 20.0 and 14.7 for Chl a, Chl b, Phe a Phe b, respectively, for concentrations in the range 0.00–1.00μgmL−1. Both MCR-ALS and Augmented PARAFAC allow one to obtain a detailed and realistic description of the analyzed samples, in terms of pure elution time, excitation and emission spectral profiles, which can be independently retrieved for every component.

Natural organic matter in urban aerosols: Comparison between water and alkaline soluble components using excitation–emission matrix fluorescence spectroscopy and multiway data analysis

Understanding the complexity of Natural Organic Matter (NOM) in atmospheric aerosols has remained an important goal for the atmospheric research community. This work employs a Parallel Factor Model (PARAFAC) with Alternating Least Squares (ALS) algorithm to decompose and further compare sets of excitation-emission matrices fluorescence spectra of Water-soluble and Alkaline-soluble Organic Matter (WSOM and ASOM, respectively), sequentially extracted from urban aerosols collected during different seasons. The PARAFAC-ALS modelling identified three components in both WSOM and ASOM, whose maximum intensities follow a clear seasonal trend and which are likely to represent the dominant fluorescent moieties in NOM from urban aerosols. The PARAFAC-ALS modelling also indicated differences between the colder and warmer seasons in the fluorescence map of one WSOM component, which contrast with the results obtained for the ASOM, where the fluorescence signatures were found to be constant along the seasons, suggesting that the ASOM may have an in situ origin.

Effect of short-time hydrothermal pretreatment of kitchen waste on biohydrogen production: Fluorescence spectroscopy coupled with parallel factor analysis

The enhancement of bio-hydrogen production from kitchen waste by a short-time hydrothermal pretreatment at different temperatures (i.e., 90°C, 120°C, 150°C and 200°C) was evaluated. The effects of temperature for the short-time hydrothermal pretreatment on kitchen waste protein conversion and dissolved organic matter characteristics were investigated in this study. A maximum bio-hydrogen yield of 81.27mL/g VS was acquired at 200°C by the short-time hydrothermal pretreatment during the anaerobic fermentative hydrogen production. Analysis of the dissolved organic matter composition showed that the protein-like peak dominated and that three fluorescent components were separated using fluorescence excitation–emission matrix spectra coupled with the parallel factor model. The maximum fluorescence intensities of protein-like components decomposed through the parallel factor analysis has a significant correlation with the raw protein concentration, showed by further correlation analysis. This directly impacted the hydrogen production ability.

Ambient modal identification of structures equipped with tuned mass dampers using parallel factor blind source separation

In this paper, a novel PARAllel FACtor (PARAFAC) decomposition based Blind Source Separation (BSS) algorithm is proposed for modal identification of structures equipped with tuned mass dampers. Tuned mass dampers (TMDs) are extremely effective vibration absorbers in tall flexible structures, but prone to get de-tuned due to accidental changes in structural properties, alteration in operating conditions, and incorrect design forecasts. Presence of closely spaced modes in structures coupled with TMDs renders output-only modal identification difficult. Over the last decade, second-order BSS algorithms have shown significant promise in the area of ambient modal identification. These methods employ joint diagonalization of covariance matrices of measurements to estimate the mixing matrix (mode shape coefficients) and sources (modal responses). Recently, PARAFAC BSS model has evolved as a powerful multi-linear algebra tool for decomposing an nth order tensor into a number of rank-1 tensors. This method is utilized in the context of modal identification in the present study. Covariance matrices of measurements at several lags are used to form a 3rd order tensor and then PARAFAC decomposition is employed to obtain the desired number of components, comprising of modal responses and the mixing matrix. The strong uniqueness properties of PARAFAC models enable direct source separation with fine spectral resolution even in cases where the number of sensor observations is less compared to the number of target modes, i.e., the underdetermined case. This capability is exploited to separate closely spaced modes of the TMDs using partial measurements, and subsequently to estimate modal parameters. The proposed method is validated using extensive numerical studies comprising of multi-degree-of-freedom simulation models equipped with TMDs, as well as with an experimental set-up.

Semi-Blind Receivers for Joint Symbol and Channel Estimation in Space-Time-Frequency MIMO-OFDM Systems

In wireless communications, increased spectral efficiency and low error rates can be achieved by means of space-time-frequency coded MIMO-OFDM systems. In this work, we consider a MIMO-OFDM transmit signal design combining space-frequency modulation with a time-varying linear precoding technique which allows spreading and multiplexing the transmitted symbols, in both space, time and frequency domains. For this system, we propose two closed-form semi-blind receivers that exploit differently the multilinear structure of the received signal, which is formulated as a nested PARAllel FACtor (PARAFAC) model. First, we devise a least squares Khatri-Rao factorization (LS-KRF) based receiver for joint channel and symbol estimation by making an efficient use of a short frame of pilot symbols. The LS-KRF receiver provides the same performance at a lower computational complexity compared to the alternating least squares (ALS) based receiver. For further reducing pilot overhead, we develop a simplified closed-form PARAFAC (S-CFP) receiver coupled with a pairing algorithm that yields an unambiguous estimation of the transmitted symbols without the need of a pilot frame. The uniqueness conditions, spectral efficiency and computational complexity of the LS-KRF and S-CFP with pairing receivers are analyzed and compared with the ALS receiver. It is shown that the S-CFP with pairing receiver has the same order of computational complexity as the ALS receiver. Meanwhile, simulation results show that our S-CFP with pairing receiver achieves the same or very similar performance of the competing receivers with extra pilot overhead at sufficiently high signal-to-noise ratio (SNR) conditions. On the other hand, it is slightly inferior to them in terms of channel estimation accuracy and bit error rate at lower SNRs.

Pyrolysis temperature-dependent release of dissolved organic carbon from plant, manure, and biorefinery wastes

Limited information is available to understand the chemical structure of biochar's labile dissolved organic carbon (DOC) fraction that will alter soil organic carbon (SOC) composition. This study utilized the high sensitivity of fluorescence excitation–emission (EEM) spectrophotometry to understand the structural changes in biochar-derived DOC as a function of (1) chemical property of feedstock and (2) pyrolysis temperature (350–800°C). Regardless of feedstock (almond shell, broiler litter, lignin, cottonseed hull, and pecan shell), low temperature (350–400°C) pyrolysis shifted EEM of hot water (80°C for 16h) extracts toward longer emission wavelengths, higher aromaticity regions. Five component parallel factor (PARAFAC) modeling of EEM afforded: (C1) (poly)phenolic pyrolysis products and other water-soluble aromatic structures similar to fulvic-like SOC; (C2–C3) aromatic structures similar to humic-like SOC having low water solubility that decomposed above 350°C; (C4) carboxyl and other transient thermochemical conversion intermediates that decomposed above 500°C; and (C5) thermally stable DOC fraction of lignin-rich biomass (pecan shell and lignin). Relative contribution of fulvic-like component exceeded that of humic-like components, and increased as a function of pyrolysis temperature. The C4 showed disproportionately high contributions to 500°C broiler litter and cottonseed hull biochars. The carboxyl-enriched DOC of biochars pyrolyzed at 350–500°C can have diverse environmental consequences, including the mobilization of heavy metals.

Passive localization of 3-D near-field sources in coexisted multiplicative noise and additive noise

By exploiting favorable characteristics of a uniform cross-array, a passive localization algorithm of narrowband sources in the spherical coordinates (azimuth, elevation and range) is proposed. Based on the properly chosen sensor outputs, we compute the third-order cyclic moment matrices, and exploit a pre-calibration technique to eliminate multiplicative noise. Then, we construct a parallel factor (PARAFAC) model, and adopt trilinear alternating least squares regression (TALS) to estimate three-dimensional (3-D) near-field parameters. The investigated algorithm is efficient in the sense that it can eliminate multiplicative noise and additive noise, provide the improved estimation accuracy, as well as avoid the parameter-pairing procedure. Simulation results are carried out to demonstrate the performance of the proposed algorithm.

Unified tensor model for space-frequency spreading-multiplexing (SFSM) MIMO communication systems

This paper presents a unified tensor model for space–frequency spreading-multiplexing (SFSM) multiple-input multiple-output (MIMO) wireless communication systems that combine space- and frequency-domain spreadings, followed by a space–frequency multiplexing. Spreading across space (transmit antennas) and frequency (subcarriers) adds resilience against deep channel fades and provides space and frequency diversities, while orthogonal space–frequency multiplexing enables multi-stream transmission. We adopt a tensor-based formulation for the proposed SFSM MIMO system that incorporates space, frequency, time, and code dimensions by means of the parallel factor model. The developed SFSM tensor model unifies the tensorial formulation of some existing multiple-access/multicarrier MIMO signaling schemes as special cases, while revealing interesting tradeoffs due to combined space, frequency, and time diversities which are of practical relevance for joint symbol-channel-code estimation. The performance of the proposed SFSM MIMO system using either a zero forcing receiver or a semi-blind tensor-based receiver is illustrated by means of computer simulation results under realistic channel and system parameters.

Enhanced Robustness of Calibration Models Using Parallel Factor (PARAFAC) Analysis with NIR Spectral Data for Non-invasive Blood Glucose Monitoring

The main cause of the low prediction accuracy in non-invasive blood glucose monitoring with near-infrared (NIR) spectroscopy is that the variations induced by the changes of the measuring system in the prediction data set are incon- sistent with those in the calibration data set. In this paper, a method to improve the robustness of the calibration models is proposed, in which the information of the matrix background is introduced as a variable into the calibration data and the three-way tensor is used to build the regression model. The idea of constructing regression models based on the hybrid algo- rithms consists of two steps. The first is to build a parallel factor (PARAFAC) model with its second-order advantage and calculate the scores and loadings. Then a multivariate linear regression (MLR) calibration model is built from the PARAFAC sample scores combined with the reference concentration values for quantification purposes. For the validation and predic- tion, the PARAFAC loadings are used to calculate the predicted scores with the validation and prediction data sets, and then the predicted concentration values can be deduced from the MLR model. The proposed method has been successfully applied to two NIR spectroscopy experiments. One is a Monte-Carlo simulation experiment of skin. The changes of the absorption coefficients and scattering coefficients of dermis are considered as the variations of the matrix background. The other is an in vitro experiment including glucose, haemoglobin and albumin solutions and the mixed composition solutions. The determi- nation coefficients and root mean square error of prediction (RMSEP) values obtained from the PARAFAC-MLR models are compared with those obtained from traditional chemometrics tools such as partial least squares (PLS). The results show that the PLS model cannot handle uncalibrated variations whereas the way of introducing the matrix background to generate ten- sor data and the regression method based on the combination of PARAFAC and MLR perform better in model robustness and prediction precision. Keywords near-infrared spectroscopy; parallel factor (PARAFAC); multivariate linear regression (MLR); Monte-Carlo; glucose; background

Microbial Transformation of Biomacromolecules in a Membrane Bioreactor: Implications for Membrane Fouling Investigation

BackgroundThe complex characteristics and unclear biological fate of biomacromolecules (BMM), including colloidal and soluble microbial products (SMP), extracellular polymeric substances (EPS) and membrane surface foulants (MSF), are crucial factors that limit our understanding of membrane fouling in membrane bioreactors (MBRs).FindingsIn this study, the microbial transformation of BMM was investigated in a lab-scale MBR by well-controlled bioassay tests. The results of experimental measurements and mathematical modeling show that SMP, EPS, and MSF had different biodegradation behaviors and kinetic models. Based on the multi-exponential G models, SMP were mainly composed of slowly biodegradable polysaccharides (PS), proteins (PN), and non-biodegradable humic substances (HS). In contrast, EPS contained a large number of readily biodegradable PN, slowly biodegradable PS and HS. MSF were dominated by slowly biodegradable PS, which had a degradation rate constant similar to that of SMP-PS, while degradation behaviors of MSF-PN and MSF-HS were much more similar to those of EPS-PN and EPS-HS, respectively. In addition, the large-molecular weight (MW) compounds (>100 kDa) in BMM were found to have a faster microbial transformation rate compared to the small-MW compounds (<5 kDa). The parallel factor (PARAFAC) modeling of three-dimensional fluorescence excitation-emission matrix (EEM) spectra showed that the tryptophan-like PN were one of the major fractions in the BMM and they were more readily biodegradable than the HS. Besides microbial mineralization, humification and hydrolysis could be viewed as two important biotransformation mechanisms of large-MW compounds during the biodegradation process.SignificanceThe results of this work can aid in tracking the origin of membrane foulants from the perspective of the biotransformation behaviors of SMP, EPS, and MSF.

Evolution of the optical properties of seawater influenced by the Deepwater Horizon oil spill in the Gulf of Mexico

The fluorescence excitation–emission matrix (EEM) technique coupled with parallel factor (PARAFAC) modeling and measurements of bulk organic carbon and other optical properties were used to characterize the oil components released from the Deepwater Horizon oil spill in the Gulf of Mexico and to examine the chemical evolution and transformation of oil in the water column. Seawater samples were collected from the Gulf of Mexico during October 2010 and October 2011, three months and fifteen months, respectively, after the oil spill was stopped. Together with previous results from samples collected during the oil spill in May/June 2010, these time series samples allow us to elucidate changes in the optical properties of dissolved organic matter (DOM) from the time of maximum oil impact to its recovery, 15 months after the spill. Although the oil had profoundly altered the optical properties of the DOM in the entire water column during the oil spill, naturally occurring DOM became predominant in surface waters by October 2010, three months after the spill. Anomalous DOM with high optical yields, however, still resided in deep waters even 15 months after the oil spill in October 2011, showing a persistent influence of the oil in deep waters. Based on fluorescence EEM data and PARAFAC modeling, three oil components and one natural humic-like DOM could be readily identified. The most prominent oil component had its maximum fluorescence intensity at Ex/Em 224/328 nm, and the other two centered on Ex/Em 264/324 and 232/346 nm, respectively. The humic-like DOM component had its wide emission peak from 390 to 460 nm over the excitation wavelength at ∼248 nm. We hypothesized that component-2 (264/324 nm) was mostly derived from photochemical degradation and the component-3 (232/346 nm) could be a degradation product from both microbial and photochemical degradation, although both C2 and C3 are subject to degradation at different rates. The oil component ratios, such as C2/C1 and C3/C1, were closely related to degradation states of oil and can be used as a sensitive index to track the fate, transport and transformation of oil in the water column.

Toward understanding the role of individual fluorescent components in DOM-metal binding

Knowledge on the function of individual fractions in dissolved organic matter (DOM) is essential for understanding the impact of DOM on metal speciation and migration. Herein, fluorescence excitation–emission matrix quenching and parallel factor (PARAFAC) analysis were adopted for bulk DOM and chemically isolated fractions from landfill leachate, i.e., humic acids (HA), fulvic acids and hydrophilic (HyI) fraction, to elucidate the role of individual fluorescent components in metal binding (Cu(II) and Cd(II)). Three components were identified by PARAFAC model, including one humic substance (HS)-like, one protein-like and one component highly correlated with the HyI fraction. Among them, the HS-like and protein-like components were responsible for Cu(II) binding, while the protein-like component was the only fraction involved in Cd(II) complexation. It was further identified that the slight quenching effect of HA fraction by Cd(II) was induced by the presence of proteinaceous materials in HA. Fluorescent substances in the HyI fraction of landfill leachate did not play as important a role as HS did. Therefore, it was suggested that the potential risk of aged leachate (more humified) as a carrier of heavy metal should not be overlooked.

Seasonal and spatial constraints of fluorophores in the midwestern Bay of Bengal by PARAFAC analysis of excitation emission matrix spectra

Dissolved organic matter (DOM) fluorescence was measured along with hydrochemical constituents and chlorophyll a of water collected monthly from the midwestern Bay of Bengal (Indian Ocean) that included the Godavari estuary and its adjoining coastal region during the period March–October 2009. By applying parallel factor (PARAFAC) modelling of the excitation emission matrix spectral data, five components – three humic (A, C, and M peaks) and two protein (B and T) were noticed. The tyrosine like component (B) is the most abundant throughout except for the estuary during monsoon when the UV humic like component (A) surpassed it. During the pre-monsoon season, the B, C and M fluorophores are enriched due to a higher bacterial decay of organic matter. The fluorescence index (FI) was higher and the humification index (HIX) lower during pre-monsoon. Similarly, the FI was higher and HIX lower for the coastal region than the estuary. The A:C ratio and A:M ratio are essentially seasonal indicators – they were <1 in the pre-monsoon and >5 during monsoon. The M:C ratio is essentially a spatial indicator, being <1 in the estuary and >1 in the coastal region during both seasons.

New Developments for the Sensitivity Estimation in Four-Way Calibration with the Quadrilinear Parallel Factor Model

Appropriate closed-form expressions are known for estimating analyte sensitivities when calibrating with one-, two-, and three-way data (vectors, matrices, and three-dimensional arrays, respectively, built with data for a group of samples). In this report, sensitivities are estimated for calibration with four-way data using the quadrilinear parallel factor (PARAFAC) model, making it possible to assess important figures of merit for method comparison or optimization. The strategy is based on the computation of the uncertainty in the fitted PARAFAC parameters through the Jacobian matrix. Extensive Monte Carlo noise addition simulations in four-way data systems having widely different overlapping situations are helpful in supporting the present approach, which was also applied to two experimental analytical systems. With this proposal, the estimation of the PARAFAC sensitivity for calibration scenarios involving three- and four-way data may be considered complete.

Spectral Characteristics in the Degradation Process of Colored Dissolved Organic Matter from Large Filamentous Algae Sources in Saltwater Lake

The indoor experimental study was on a large common species of filamentous green algae in the North Lake beach of Chongming Island - bristles algae liquid produced by algae growth during the photodegradation of colored dissolved organic matter (CDOM) spectral characteristics change. Absorption coefficient a (355) in the degradation experiment was reduced 27.44% , however , in the control experiment , a (355) was only 1.61% lower. In the photodegradation process of degradation experiments and control experiments, the values of M and S were increased to the different extent and the increased value of the former was higher than the latter. By using three-dimensional fluorescence spectroscopy (EEMS), the detected peak value of protein-and humic fluorescence in control experiments was little changed, while significantly increased in the degradation experiments. By the analysis of parallel factor model (PARAFAC) combined with three-dimensional fluorescence excitation - emission matrix data (EEMS), the three-component CDOM: C1, C2, C3, respectively were relative to humic fluorescence peaks A, C, and protein-like fluorescence B. In the degradation experiments, small changes were in the components of C1, however the components of C2, C3 compared with those in the control group were increased by nearly 2 times. All results showed that the photodegradation process of bristles algae liquid can cause changes of the CDOM in the structure and content, easier to produce small molecules and weak aromatic CDOM.

Diagnosis of early stage nasopharyngeal carcinoma using ultraviolet autofluorescence excitation–emission matrix spectroscopy and parallel factor analysis

We report the diagnostic ability of ultraviolet (UV)-excited autofluorescence (AF) excitation-emission matrix (EEM) spectroscopy associated with parallel factor (PARAFAC) analysis for differentiating cancer from normal nasopharyngeal tissue. A bifurcated fiber-optic probe coupled with an EEM system was used to acquire tissue AF EEMs using excitation wavelengths between 260 and 400 nm, and emission collection between 280 and 500 nm. A total of 152 AF EEM landscapes were acquired from 13 normal and 16 nasopharyngeal carcinoma (NPC) thawed ex vivo tissue samples from 23 patients. PARAFAC was introduced for curve resolution of individual AF EEM landscapes associated with the endogenous tissue constituents. The significant factors were further fed to a support vector machine (SVM) and cross-validated to construct diagnostic algorithms. Both the EEM intensity landscapes and the PARAFAC model revealed tryptophan, collagen, and elastin to be the three major endogenous fluorophores responsible for the AF signal from normal and NPC tissues. The EEM intensity distribution and PARAFAC factors suggest an increase of tryptophan and a decrease of collagen and elastin in NPC tissues compared to the normal. The classification results obtained from the PARAFAC-SVM modeling yielded a diagnostic accuracy of 94.7% (sensitivity of 95.0% (76/80); specificity of 94.4% (68/72)) for normal and NPC tissue differentiation. This study suggests that UV-excited AF EEM spectroscopy integrated with PARAFAC algorithms has the potential to provide clinical diagnostics of early onset and progression of NPC.