March 2012
Volume 53, Issue 3
Free
Immunology and Microbiology  |   March 2012
Differentiation between Infectious and Noninfectious Ulcerative Keratitis by Raman Spectra of Human Teardrops: A Pilot Study
Author Affiliations & Notes
  • Ming-Tse Kuo
    From the Department of Biomedical Engineering,
    the Department of Ophthalmology, Kaohsiung Chang Gung Memorial Hospital and Chang Gung University College of Medicine, Kaohsiung, Taiwan; and
  • Chi-Chang Lin
    the Department of Chemical and Materials Engineering, Tunghai University, Taichung, Taiwan.
  • Hsin-Yu Liu
    From the Department of Biomedical Engineering,
  • Mei-Yu Yang
    From the Department of Biomedical Engineering,
  • Hsien-Chang Chang
    From the Department of Biomedical Engineering,
    Center for Micro/Nano Science and Technology, and
    Medical Device Innovation Center, National Cheng Kung University, Tainan, Taiwan;
  • Corresponding author: Hsien-Chang Chang, Department of Biomedical Engineering, National Cheng Kung University, No. 1, University Road, Tainan 70101, Taiwan, R.O.C.; [email protected]
Investigative Ophthalmology & Visual Science March 2012, Vol.53, 1436-1444. doi:https://doi.org/10.1167/iovs.11-7923
  • Views
  • PDF
  • Share
  • Tools
    • Alerts
      ×
      This feature is available to authenticated users only.
      Sign In or Create an Account ×
    • Get Citation

      Ming-Tse Kuo, Chi-Chang Lin, Hsin-Yu Liu, Mei-Yu Yang, Hsien-Chang Chang; Differentiation between Infectious and Noninfectious Ulcerative Keratitis by Raman Spectra of Human Teardrops: A Pilot Study. Invest. Ophthalmol. Vis. Sci. 2012;53(3):1436-1444. https://doi.org/10.1167/iovs.11-7923.

      Download citation file:


      © ARVO (1962-2015); The Authors (2016-present)

      ×
  • Supplements
Abstract

Purpose.: The aim of this study was to use Raman spectra of tears to differentiate between infectious and noninfectious ulcerative keratitis.

Methods.: Raman microspectroscopy was applied using the drop-coating deposition method on Ti/Au-coated glass slides to obtain sample spectra from different human tear groups, including tears from healthy subjects and from patients with infectious and noninfectious ulcerative keratitis. By comparing the difference spectra of the groups, the authors identified local Raman features useful for differentiation of ulcerative keratitis. Principal components (PCs) of normal tears were used as affined spectral coordinates. After performing projections of Raman spectra of both infectious and noninfectious tear samples, the authors compared the two groups to identify global spectral parameters with differential statistical significance.

Results.: Differentiation between infectious and noninfectious ulcerative keratitis might be made directly through observation of the normalized tear Raman spectra or the transformed principal scores. Spectral segments with differential statistical significance included 878∼888 cm−1, 885∼888 cm−1, 945∼993 cm−1, 1007∼1015 cm−1, 1074∼1100 cm−1, 1090∼1094 cm−1, 1096∼1099 cm−1, 1386∼1403 cm−1, 1463∼1469 cm−1, 1469∼1473 cm−1, 1557∼1563 cm−1, 1584∼1588 cm−1, and 1614∼1621 cm−1. There were two PCs with statistically significant differences for the two groups of ulcerative keratitis, PC1 (P = 0.01) and PC2 (P = 0.05).

Conclusions.: This novel approach using the analysis of Raman spectra of teardrop samples for differentiation of ulcerative keratitis demonstrates the potential application of Raman microspectroscopy for clinical practice. This technology should complement the conventional cytological method for rapid diagnosis in the clinician's office.

Infectious ulcerative keratitis (IUK), which manifests as painful eye redness and a cloudy cornea caused by infectious pathogens, 1 4 is a vision-threatening disease worldwide. 5 9 Noninfectious ulcerative keratitis (NIUK), 10,11 which has a clinical presentation similar to that of IUK but no known infectious cause, 12 is a diverse disease associated with systemic disorders and peripheral corneal invasion. 13,14 The development of corneal ulceration in systemic autoimmune disease may represent the progression of a life-threatening disease. 14,15 Contact lens wear 16,17 and trauma 12 are common risk factors 18,19 for IUK and NIUK. Antimicrobial agents are used to treat IUK, 20 23 whereas immunosuppressive agents are used to treat NIUK. 11,24 26 Delayed diagnosis or misdiagnosis is possible because it can be difficult to differentiate between IUK and NIUK, 27 29 and surgical management may be needed for both IUK and NIUK. 30,31  
Rapid diagnostic aid for IUK is important to reduce the complications caused by delayed diagnosis. However, few methods can diagnose IUK within the 2-hour time frame required in the ophthalmologist's clinic. 32,33 Cytological diagnosis uses light microscopy to observe stained samples from corneal scraping or impression techniques. 34,35 However, the detection limit of bacterial density (≥105 counts/mL) 36,37 of this method may cause sensitivity to be reduced to 40% and yield a false-negative rate of up to 60%. 38 In vivo confocal microscopy is another rapid and less invasive technique for the diagnosis of IUK 39,40 ; however, its discriminative power is limited in relatively large microorganisms, such as Acanthamoeba species and large fungi. 41,42 Although microbial culture is the gold standard for the diagnosis of IUK, the incubation time is at least 2 days for rapidly growing bacteria and up to 2 to 4 weeks for viruses, fungi, and mycobacteria. 43 Polymerase chain reaction (PCR)-based methods can be considered a new standard for IUK diagnosis, which might shorten the diagnostic time to 4 to 8 hours, 44,45 but they still do not suit the need for more rapid diagnosis in a clinic. 
Teardrops, which contain shed microbes and their metabolites, 46 49 may be a good target by a scraping-free and less invasive collection way to differentiate IUK from NIUK. Changes in the intrinsic tear components and their concentrations may also reflect the defense mechanisms of the ocular surface against microbes or other stimuli. 48,50 58 Lysozyme, a protective protein in tears, has been used to distinguish patients with herpes simplex virus (HSV)–caused keratitis from healthy subjects. 59 Levels of leukotriene B4 and platelet-activating factor in the tears of patients with contact lens–induced peripheral ulcers were also reported to be higher than those in the tears from the control subjects. 58  
Raman spectra were first discovered by Raman and Krishnan in 1928. 60 Raman microspectroscopy is a novel technology that applies the inelastic scattering of light to qualitatively and quantitatively detect the molecular composition of matter in real time. 61 Raman microspectroscopy can detect the scattering photons and provides molecular information through changes in the levels of vibrating energy following the monochromic laser photons in the visible light spectrum, impacting the detected sample. Many studies have shown its potential application in different biomedical investigations, 62 73 including distinguishing microbes and their metabolites. 74 80 Raman microspectroscopy was recently determined to be a powerful tool for detecting tear components. Zhang et al. 81 have shown that a high efficiency of drop-coating deposition Raman (DCDR) can be achieved by the deposition of aqueous samples on substrates with specific surface properties. Filik and Stone 82,83 demonstrated that a Raman signal with a high signal-to-noise ratio was obtained from 1.5 μL dried teardrop for the detection of major tear components using the DCDR technique. 
Principal component analysis (PCA) is a multivariate statistical technique used in almost all scientific disciplines. 84,85 PCA analyzes a data matrix that represents the observations recorded by several dependent or intercorrelated variables. Its goal is to extract dominant information from the complicated data matrix and to express the features as a set of new orthogonal variables called principal components. In other words, PCA greatly concentrates many less informative dependent variables (dimension n) of a data matrix to few more informative principal components (dimension m), which may represent the important features (number of features, m ≪ number of variables, n) of this matrix. Therefore, PCA is widely used in feature extraction. 86 88  
The aim of this study was to discover the differentiation potential of human tear DCDR to identify optical fingerprints to further aid the differentiation between IUK and NIUK; this will be useful in the future verification study for clinical application. 
Methods
Patients
Tear samples were prospectively collected from August 1, 2009, to December 1, 2011, from patients in the Chang Gung Memorial Hospital (CGMH), Kaohsiung Medical Center, including tears from healthy eyes (Norm group) and tears from diseased eyes with acute ulcerative keratitis (UK). All procedures involving human subjects adhered to the Declaration of Helsinki and were approved by the Committee of Medical Ethics and Human Experiments of CGMH. Informed consent was obtained from each subject in the Kaohsiung Medical Center of CGMH. 
Collection of Clinical Data
The eyes of the Norm group were healthy on the ocular surface based on biomicroscopy evaluation. For each patient, local and systemic risk factors of UK were recorded after considering the patient's history and findings from ocular examination. Risk factors were contact lens wear, corneal trauma (foreign body trauma, previous corneal surgery), systemic disease (such as autoimmunity, diabetes, and immunosuppression), ocular surface and corneal diseases (such as blepharitis, trichiasis, limbal stem cell deficiency, and allergic keratoconjunctivitis), and previous microbial keratitis (such as herpetic keratitis and bacterial keratitis). 89 We included patients with UK, whose clinical presentation included corneal epithelial defects with stromal inflammation and infiltrates. Patients younger than 9 years of age were excluded, as were patients who had received topical fluorescence staining before the collection of tears. 
Definition of Ulcerative Keratitis with or without Infection
We classified UK into two groups: infectious UK (IUK group) and noninfectious UK (NIUK group). UK with infection is defined as any positive finding identified in the microbial survey protocol for confirmation of infectious keratitis for originally suspected infectious keratitis. This protocol includes cytological evaluation, microbial culture, and HSV PCR of the corneal scraping samples. For cytological evaluation, the Gram and acid-fast stains were used for immediate microscopic evaluation. If the diameter or the maximal length of the ulcer was <2 mm, the cytological evaluation was omitted to preserve the samples for microbial culture or HSV PCR (in-house real-time PCR for HSV types 1 and 2). Routine microbial culture included aerobic culture (blood agar and chocolate blood agar), anaerobic culture (CDC anaerobe 5% sheep blood agar, phenylethyl alcohol blood agar, and bacteroides bile esculin agar), mycobacterial culture (Lowenstein-Jensen slant), and fungal culture (Sabouraud's dextrose agar). On any specific suspicion of atypical corneal ulceration at the initial visit, we also performed Acanthamoeba culture (nonnutrient agar with Escherichia coli overlay) and HSV PCR. If treatment took an unexpected clinical course, we repeated the standard microbial survey and added Acanthamoeba culture or HSV PCR. UK without infection is defined as no positive findings by the microbial confirmation protocol for presumed microbial keratitis at the initial visit, smooth tapering of antimicrobial agents, and no occurrence of unexpected events until smoothing of the ocular surface. 
Collection of Human Tears
To minimize interference from topical compounds in case of pretreated patients, tear collection was performed at least 1 hour after topical medications were stopped. Tears were gently collected from all participants under 16× magnification with a slit lamp using disinfected microcapillary tubes (Hirschmann Laborgeräte GmbH & Co. KG, Eberstadt, Germany) with modified head curvature. Tear samples (5–10 μL) were collected and stored at 4°C immediately before spectral acquisition using Raman microspectroscopy within 2 weeks. 
Dried Teardrop Preparation for Raman Spectral Acquisition
We applied Raman microspectroscopy (Thermo Fisher Scientific Inc., Waltham, MA) based on the DCDR method 83,90 to Ti/Au-coated glass slides with slight modification, as described in our previous study. 91 In brief, 1.5 μL tear sample was dropped onto a clean glass slide coated with 30/150 nm Ti/Au. The slide was placed in a closed chamber and allowed to dry at room temperature with 48% relative humidity. Raman spectra in the central zone of the dried teardrop were sampled by a line-mapping procedure. Poorly formed dried teardrops resulting from inappropriate pipette control on the Ti/Au-coated glass slide were excluded from further spectral collection. Raman spectra of the tear samples were collected under the following experimental conditions: 633 nm He-Ne laser excitation, laser power at the sample position of approximately 7 mW, and a 50× objective lens with a connected charge-coupled device for scattered light collection. The Raman shift was registered using a signal of 520 cm−1 generated from a silicon wafer. A laser exposure time of 10 seconds with a signal accumulation of three times was applied for each spectral acquisition point. Postprocessing of the spectra was conducted as described elsewhere, 92,93 with some modifications. 91 In brief, the fluorescence background was removed by fitting and subtracting a fifth-order polynomial to the full spectrum. A collection of 10 spectra in the central zone was obtained for each teardrop. Each spectrum was normalized using the intense band at 1002 cm−1, and the mean was centered to get a zero mean for spectral pattern analysis and statistics. 
Statistical Analysis
For each subject, 10 Raman spectra in the central zone of a dried teardrop were collected for analysis. We collected Raman signals in the range of 700 to 1800 cm−1 for analysis and used a spreadsheet (Excel 2007; Microsoft Corporation, Redmond, WA) as a graphic tool. Mean spectra of tear samples from the three groups (NIUK-Norm, IUK-Norm, and IUK-NIUK) and difference spectra were visually compared to determine the candidate parameters of differential Raman shifts and those of differential segments of the Raman spectra. Integrals of candidate segments were formatted as new parameters for differential purposes. These local spectral parameters were then univariantly tested with the Mann-Whitney U test for the two diseased groups. We used technical computing software (MatLab 2008a; The MathWorks Inc., Natick, MA) as the multivariable analytical tool for PCA to extract the global spectral features of tears from the normal tear group. 83,88,94 The determined principal vectors were used as the transfer function to determine the principal component (PC) scores for each spectrum of each subject. The first few PCs for which the relevance cumulated to 80% were chosen as candidate global parameters. The mean scores of these candidates in the two diseased groups were compared univariantly by one-tailed t-test to evaluate their statistical significance. Analysis of subgroups in IUK samples were based on candidate global parameters with statistical significance. 
Results
Clinical Data Analysis
There were 30 subjects in the Norm tear group, 26 patients in the IUK group, and 26 patients in the NIUK group (Table 1). There was no statistically significant difference with respect to sex (χ2 = χ2 test with Yates correction; P = 0.67), age (one-way ANOVA, P = 0.68), or laterality of the eye (χ2; P = 0.85) among the three groups. Comparison of the two groups indicated that contact lens wear and corneal trauma were the most common risk factors for both. Although there was no statistical significant difference for the overall frequencies of these risk factors (χ2; P = 0.78), ocular surface diseases were more frequent in the NIUK group and previous UK was more frequent in the IUK group. Because CGMH is a tertiary medical center, several patients had received antimicrobial medication before they participated in this study. However, there was no statistically significant difference for both groups (χ2; P = 0.97). Analyzing the morphologic factors for UK indicated a statistically significant difference (χ2; P = 0.03) for ulcer boundary pattern because the IUK group had more elliptical ulceration than the NIUK group. There was a tendency for larger ulcer area in the IUK group, but the difference did not reach the statistical significance (one-tailed t-test; P = 0.09). There was no statistically significant difference for depth (χ2; P = 0.35) or location (χ2; P = 0.27). The identified pathogens and risk factors for the IUK group are summarized in Supplementary Table S1
Table 1.
 
Clinical Data of the Three Participant Groups
Table 1.
 
Clinical Data of the Three Participant Groups
Clinical Features Norm Group IUK Group NIUK Group
Case number 30 26 26
Age, y 42.7 ± 13.5 46.5 ± 17.8 46.1 ± 21.3
Sex, male/female 15/15 9/17 12/14
Laterality, OD/OS 16/14 11/15 12/14
Risk factors for keratitis
    Contact lens wear 9 7
    Corneal trauma 10 10
    Systemic disease 2 2
    Ocular surface and corneal disease 2 6
    Previous ulcerative keratitis 3 1
Topical antimicrobials before visit
    No 6 5
    Yes, ≤3 days 8 8
    Yes, >3 days 12 13
Boundary pattern of ulcer
    Near elliptical shape 19 10
    Nonelliptical shape 7 16
Area of ulcer, mm2 10.2 ± 9.3 6.5 ± 10.8
Depth of ulcer (anterior one-third/posterior two-thirds) 17/9 21/5
Location of ulcer (central/paracentral/peripheral) 12/11/3 8/9/9
Comparison of Raman Spectra of Samples from the Normal Tear Group, NIUK Group, and IUK Group to Select Local Features in Spectra for IUK Identification
The mean Raman spectra from the Norm, NIUK, and IUK tear groups showed similar patterns, but several different features can be pointed out clearly (Fig. 1a). There were two small peaks on the segment 753∼757 cm−1 in the IUK group, whereas there was only one small peak on the same band in the NIUK and Norm groups. Upward deviation was revealed from the end of segment 1074∼1100 cm−1 to the start of segment 1176∼1182 cm−1 in the IUK group, whereas negative deviation was observed on the same region in the other two groups. There were two characteristic peaks on the segments 1227∼1241 cm−1 and 1275∼1291 cm−1 in the IUK group. Although there were other local features among the three groups, we did not describe them one by one to avoid redundancy. 
Figure 1.
 
Mean Raman spectra of tear samples with 95% confidence intervals of different tear groups, including healthy subjects (300 spectra of 30 subjects), noninfectious keratitis (260 spectra of 26 patients), and infectious keratitis (260 spectra of 26 patients) (a). Difference in the Raman spectra of tear samples with 95% confidence intervals of different tear groups (b).
Figure 1.
 
Mean Raman spectra of tear samples with 95% confidence intervals of different tear groups, including healthy subjects (300 spectra of 30 subjects), noninfectious keratitis (260 spectra of 26 patients), and infectious keratitis (260 spectra of 26 patients) (a). Difference in the Raman spectra of tear samples with 95% confidence intervals of different tear groups (b).
The patterns of difference spectra in Figure 1b were similar between the IUK-Norm and IUK-Norm spectra, whereas the oscillating pattern around the baseline was shown in the NIUK-Norm spectrum. These difference spectra indicated the IUK group is clearly different, either from the NIUK group or from the Norm group. Based on Figure 1b, we plotted the red and blue vertical bands for all spectral figures: the red vertical bands indicate that there were visible differences in any two groups. The blue vertical bands indicate that there were visible differences only between the IUK group and the NIUK group and between the IUK group and the Norm group. We statistically tested these selected local features between the IUK group and the NIUK group and found many features with significant differences (Supplementary Fig. S1). The tested segments with differential potential included 878∼888 cm−1, 885∼888 cm−1, 945∼993 cm−1, 1007∼1015 cm−1, 1074∼1100 cm−1, 1090∼1094 cm−1, 1096∼1099 cm−1, 1386∼1403 cm−1, 1463∼1469 cm−1, 1469∼1473 cm−1, 1557∼1563 cm−1, 1584∼1588 cm−1, and 1614∼1621 cm−1
Extraction of Global Features of Raman Spectra from Normal Tears by PC Transformation
Ten Raman spectra in the central zone of a dried teardrop for each subject in the normal control group were collected (300 spectra sampled from 30 normal teardrops) to establish the transfer function of Raman spectra of tear samples (Fig. 2a). Because the relevance cumulated from PC1 to PC6 was approximately 80%, as indicated in Figure 2b, we recognized the six new variables as global features of Raman spectra of tear samples. The six principal vectors of the Norm tear group served as candidate transfer functions for each Raman spectrum in all tear groups. The transformed scores of the six PCs were used for the subsequent comparative analysis. 
Figure 2.
 
PC transformation of Raman spectra of tear samples is based on the Norm group. The principal vectors of Norm serve as a set of transfer functions to extract global tear features for different tear groups (a). PC1 to PC6 had a cumulated relevance of approximately 80% for the Norm group, and therefore, serve as global features of Raman spectra of tear samples (b).
Figure 2.
 
PC transformation of Raman spectra of tear samples is based on the Norm group. The principal vectors of Norm serve as a set of transfer functions to extract global tear features for different tear groups (a). PC1 to PC6 had a cumulated relevance of approximately 80% for the Norm group, and therefore, serve as global features of Raman spectra of tear samples (b).
Comparison of Infectious and Noninfectious UK by Raman PCs of Tear Samples
After transformation of the mean spectra (N = 10 for each case) to the PC domain, we compared the Norm-based principal scores of the first six PCs between the IUK and NIUK groups (Fig. 3). There were two PCs with statistically significant differences for the two groups, PC1 (P = 0.01, power = 0.98) and PC2 (P = 0.05, power = 0.98). There was a greater trend in the IUK group than in the NIUK group for PC4 (P = 0.09, power = 0.87), but there was no statistically significant difference for PC3, PC5, and PC6. We focused further on the three potential PCs (corresponding principal vectors shown in Supplementary Fig. S2) to analyze the subgroups in the IUK group. 
Figure 3.
 
Comparison of the scores of the Norm-based PCs between IUK (red block) and NIUK (black block). The principal scores were acquired on the basis of the transformation system of normal teardrops. Each box is constructed by five parameters, including the median (Q2), lower, and upper quartiles (Q1, Q3), and lowest and highest data (Q1 − 1.5 × [Q3 − Q1], Q3 + 1.5 × [Q3 − Q1]). One-tailed t-test, *P < 0.05. + Power > 0.8.
Figure 3.
 
Comparison of the scores of the Norm-based PCs between IUK (red block) and NIUK (black block). The principal scores were acquired on the basis of the transformation system of normal teardrops. Each box is constructed by five parameters, including the median (Q2), lower, and upper quartiles (Q1, Q3), and lowest and highest data (Q1 − 1.5 × [Q3 − Q1], Q3 + 1.5 × [Q3 − Q1]). One-tailed t-test, *P < 0.05. + Power > 0.8.
Estimation of Mean Scores of Significant PCs for Subgroups of IUK
In Figures 4a, 4c, and 4e, indicate the score difference in the selected PCs for the classification of the pathogens in this study. Because there were only two cases of Acanthamoeba spp. and two cases of HSV, we focused on determining score deviations from the Norm group in the fungal and bacterial groups. For the PC1 and PC4 scores, there were stronger positive deviations from the normal mean score (zero score) in the bacterial and fungal groups. In contrast, there was no apparent deviation from the zero score in the bacterial and fungal groups for the PC2 score. For plots Figures 4b, 4d, and 4f, we focused on the subgroup analysis for the bacteria group. There were similar positive deviations for PC1 and PC4 scores for the Gram-positive cocci (GPC) group and Gram-negative bacilli (GNB) group. For the PC2 score, both the GPC and GNB groups did not show apparent deviation from the zero score. Therefore, both the PC1 and the PC4 scores might be useful indicators to differentiate bacterial and fungal tears from normal tears. Joining the result in the differentiation between the IUK group and the NIUK group, the PC1 score might be a bifunctional indicator to differentiate IUK tears either from NIUK tears or from normal tears. 
Figure 4.
 
Comparison of 95% confidence intervals of the significant scores between the IUK and Norm groups. Exclusive comparisons among different microbial subgroups of IUK for the scores of PC1 (a), PC2 (c), and PC4 (e). Inclusive comparisons among overlapping bacterial subgroups for the scores of PC1 (b), PC2 (d), and PC4 (f).
Figure 4.
 
Comparison of 95% confidence intervals of the significant scores between the IUK and Norm groups. Exclusive comparisons among different microbial subgroups of IUK for the scores of PC1 (a), PC2 (c), and PC4 (e). Inclusive comparisons among overlapping bacterial subgroups for the scores of PC1 (b), PC2 (d), and PC4 (f).
Discussion
It is critical to determine whether UK is caused by infection. In addition to different management strategies for IUK and NIUK, a rapid diagnosis to minimize ulcerative complications such as central, deep corneal involvement, scarring severity, and irregular astigmatism is critical to avoid vision loss. Conventional cytological examination is quick but limited by microbe levels and invasiveness in the tissue scrapings for use in routine clinical practice. In this study, we propose a novel noninvasive approach using analysis of Raman spectra of dried teardrops to complement the cytological approach. The differentiation between IUK and NIUK might be directly made through observation of the normalized tear Raman spectra or the transformed principal scores. Global spectral parameters of the tear samples were transformed from PCA, and the local spectral parameters were selected from difference spectra by visualization methods. We found some of these candidate parameters had statistical differences between the two disease groups. This approach may be a good adjunct diagnostic technique to promote diagnostic quality for UK. 
Willemse-Erix et al. 95 showed that Raman spectroscopy is a real-time typing method for classifying clinical isolates of methicillin-resistant Staphylococcus aureus with an accuracy of up to 95%. A 1.5-μL teardrop sample from a healthy human subject is sufficient for Raman signal detection of major tear components with a high signal-to-noise ratio and for reproducibility, as shown by Filik and Stone. 83 Reyes-Goddard et al. 96 demonstrated the discrimination potential of Raman spectra of tear samples for HSV by using a synthetic tear model on silver mirror reaction glass slides and gold thin film. The mean diagnostic sensitivity and specificity were estimated between 75% and 80%. Rossi et al. 92 demonstrated that Raman spectroscopy can be used to differentiate endophthalmitis from uveitis in a rabbit model. In adjoining principal components PC3 and PC4 using Mahalanobis distance as a discriminator, the diagnostic model showed sensitivity of 89%, specificity of 100%, and accuracy of 92% by a cross-validation procedure. In our study, we used gold thin film as a substrate for a 1.5-μL dried teardrop to differentiate IUK from NIUK by Raman spectroscopy, which supported the potential clinical application of this optical fingerprinting technology. We identified a set of global spectral features for the teardrop samples based on principal components and several local spectral features with differential potential shown in statistics. 
This study focused on the discovery of clinical application of Raman spectra analysis of tear samples to differentiate UK with and without infection. We hypothesized that several low-concentration products may be released because of interactions between tears and microbes, which may be different between IUK and NIUK. Although there might be microbial involvement in the pathogenesis of NIUK, 97,98 the quorum-sensing effect likely dominates in IUK. 99 102  
There are some limitations to this pilot study. The number of cases may be too few; hence, some spectral parametric candidates might not show statistical significance. We objectively defined IUK and NIUK based on the strict results of a conventional microbial workup protocol. However, presumed microbially driven UK is often only culture-positive in 40% to 60% of cases. 7,8 UK without infection as defined might be caused by infectious entities, such as those seen in a case report of microbial keratitis initially diagnosed as a contact lens peripheral ulcer. 27 This study took place at a tertiary medical center; only approximately one-fourth of the patients had not yet received treatment before visiting our hospital. Some patients in the NIUK group might initially have had microbial infection, which might make the validity in clinical human tears lower than those shown in synthetic tears 96 and in animal vitreous fluid 92 for differentiation. In spite of the natural limitations mentioned, by combining concepts from quorum sensing with those from flora adaptation on the ocular surface, 100,103 the true classification of IUK and NIUK might overlap in the same spectrum. The summation influence from different arms and different levels controlled by microbial cross-talk, host factors, and premedications determines UK of infectious or noninfectious status. The value of differentiation into the two groups may be a useful aid for diagnosis of different stages of microbial contact and for the rapid justification of systemic survey of life-threatening disorders. As such, Raman spectra analysis of teardrops may improve diagnostic quality, rather than replace cytology, culture techniques, and PCR-based methods. In the future, this technique may be further developed as a diagnostic tool for pathogens through the accumulation of more cases with the same infectious entities, such as Pseudomonas keratitis or Staphylococcus keratitis. 
In conclusion, this novel approach of Raman spectra analysis of tear samples for differentiation of UK showed the potential application of Raman microspectroscopy for clinical practice. We suggest that the resolution of tear Raman spectra for pathogenic diagnosis will be gradually improved in the near future through the study of more cases in the IUK subgroups. 
Supplementary Materials
Figure sf01, PDF - Figure sf01, PDF 
Table st1, PDF - Table st1, PDF 
Table st2, PDF - Table st2, PDF 
Footnotes
 Supported by Chang Gung Research Proposal Grant CMRPG880701, National Science Council Grants NSC 99-2218-E-029-004 and NSC99-2314-B-182A-030-MY3, National Medical Research Proposal of Chang Gung Memorial Hospital Grant NMRPG896031, and Multidisciplinary Center of Excellence for Clinical Trial and Research Grant DOH100-TD-B-111-002 from the Department of Health, Executive Yuan, Taiwan.
Footnotes
 Disclosure: M.-T. Kuo, None; C.-C. Lin, None; H.-Y. Liu, None; M.-Y. Yang, None; H.-C. Chang, None
The authors thank the Ministry of Education of Taiwan under the NCKU Project of Promoting Academic Excellence and Developing World Class Research Centers (R017) for supplying Raman spectroscopy equipment, and they thank Tsung-Chain Chang for his kind discussion. 
References
Branco BC Gaudio PA Margolis TP . Epidemiology and molecular analysis of herpes simplex keratitis requiring primary penetrating keratoplasty. Br J Ophthalmol. 2004;88:1285–1288. [CrossRef] [PubMed]
Bourcier T Thomas F Borderie V Chaumeil C Laroche L . Bacterial keratitis: predisposing factors, clinical and microbiological review of 300 cases. Br J Ophthalmol. 2003;87:834–838. [CrossRef] [PubMed]
Shukla PK Kumar M Keshava GBS . Mycotic keratitis: an overview of diagnosis and therapy. Mycoses. 2008;51:183–199. [CrossRef] [PubMed]
McAllum P Bahar I Kaiserman I Srinivasan S Slomovic A Rootman D . Temporal and seasonal trends in Acanthamoeba keratitis. Cornea. 2009;28:7–10. [CrossRef] [PubMed]
Varaprasathan G Miller K Lietman T . Trends in the etiology of infectious corneal ulcers at the F. I. Proctor Foundation. Cornea. 2004;23:360–364. [CrossRef] [PubMed]
Green M Apel A Stapleton F . Risk factors and causative organisms in microbial keratitis. Cornea. 2008;27:22–27. [CrossRef] [PubMed]
Toshida H Kogure N Inoue N Murakami A . Trends in microbial keratitis in Japan. Eye Contact Lens. 2007;33:70–73. [CrossRef] [PubMed]
Fong CF Tseng CH Hu FR Wang IJ Chen WL Hou YC . Clinical characteristics of microbial keratitis in a university hospital in Taiwan. Am J Ophthalmol. 2004;137:329–336. [CrossRef] [PubMed]
Hsiao CH Yeung L Ma DHK . Pediatric microbial keratitis in Taiwanese children: a review of hospital cases. Arch Ophthalmol. 2007;125:603–609. [CrossRef] [PubMed]
Kigasawa K Kato K Akatsuka I Tanemoto K Ishikawa H . Eighteen-year review of noninfectious corneal ulcers. Jpn J Clin Ophthalmol. 1998;52:1841–1846.
Donzis PB Mondino BJ . Management of noninfectious corneal ulcers. Surv Ophthalmol. 1987;32:94–110. [CrossRef] [PubMed]
Diamant JI Abbott RL . Surgical management of infectious and noninfectious keratitis. Int Ophthalmol Clin. 1998;38:197–217. [CrossRef] [PubMed]
Messmer EM Foster CS . Vasculitic peripheral ulcerative keratitis. Surv Ophthalmol. 1999;43:379–396. [CrossRef] [PubMed]
Ladas JG Mondino BJ . Systemic disorders associated with peripheral corneal ulceration. Curr Opin Ophthalmol. 2000;11:468–471. [CrossRef] [PubMed]
Sainz de la Maza M Foster CS . Peripheral ulcerative keratitis and malignancies. Cornea. 1994;13:364–367. [CrossRef] [PubMed]
Levy B McNamara N Corzine J Abbott RL . Prospective trial of daily and extended wear disposable contact lenses. Cornea. 1997;16:274–276. [CrossRef] [PubMed]
Stapleton F Dart JK Minassian D . Risk factors with contact lens related suppurative keratitis. Eye Contact Lens. 1993;19:204–210.
Musch DC Sugar A Meyer RF . Demographic and predisposing factors in corneal ulceration. Arch Ophthalmol. 1983;101:1545–1548. [CrossRef] [PubMed]
Erie JC Nevitt MP Hodge DO Ballard DJ . Incidence of ulcerative keratitis in a defined population from 1950 through 1988. Arch Ophthalmol. 1993;111:1665–1671. [CrossRef] [PubMed]
Collum LM Logan P McAuliffe Curtin D Hung SO Patterson A Rees PJ . Randomised double-blind trial of acyclovir (Zovirax) and adenine arabinoside in herpes simplex amoeboid corneal ulceration. Br J Ophthalmol. 1985;69:847–850. [CrossRef] [PubMed]
Gangopadhyay N Daniell M Weih L Taylor HR . Fluoroquinolone and fortified antibiotics for treating bacterial corneal ulcers. Br J Ophthalmol. 2000;84:378–384. [CrossRef] [PubMed]
Seal DV . Acanthamoeba keratitis update—incidence, molecular epidemiology and new drugs for treatment. Eye. 2003;17:893–905. [CrossRef] [PubMed]
Kalavathy CM Parmar P Kaliamurthy J . Comparison of topical itraconazole 1% with topical natamycin 5% for the treatment of filamentous fungal keratitis. Cornea. 2005;24:449–452. [CrossRef] [PubMed]
Clewes AR Dawson JK Kaye S Bucknall RC . Peripheral ulcerative keratitis in rheumatoid arthritis: successful use of intravenous cyclophosphamide and comparison of clinical and serological characteristics. Ann Rheum Dis. 2005;64:961–962. [CrossRef] [PubMed]
Galor A Jabs DA Leder HA . Comparison of antimetabolite drugs as corticosteroid-sparing therapy for noninfectious ocular inflammation. Ophthalmology. 2008;115:1826–1832. [CrossRef] [PubMed]
Tauber J Sainz de la Maza M Hoang-Xuan T Foster CS . An analysis of therapeutic decision making regarding immunosuppressive chemotherapy for peripheral ulcerative keratitis. Cornea. 1990;9:66–73. [CrossRef] [PubMed]
Diec J Carnt N Tilia D . Prompt diagnosis and treatment of microbial keratitis in a daily wear lens. Optom Vis Sci. 2009;86:E904–E907. [CrossRef] [PubMed]
Aasuri MK Venkata N Kumar VM . Differential diagnosis of microbial keratitis and contact lens-induced peripheral ulcer. Eye Contact Lens. 2003;29:S60–S62. [CrossRef] [PubMed]
Srinivasan M Mascarenhas J Prashanth CN . Distinguishing infective versus noninfective keratitis. Indian J Ophthalmol. 2008;56:203–207. [CrossRef] [PubMed]
Kim HK Park HS . Fibrin glue-assisted augmented amniotic membrane transplantation for the treatment of large noninfectious corneal perforations. Cornea. 2009;28:170–176. [CrossRef] [PubMed]
Portnoy SL Insler MS Kaufman HE . Surgical management of corneal ulceration and perforation. Surv Ophthalmol. 1989;34:47–58. [CrossRef] [PubMed]
McDonnell PJ Nobe J Gauderman WJ Lee P Aiello A Trousdale M . Community care of corneal ulcers. Am J Ophthalmol. 1992 114:531–538. [CrossRef] [PubMed]
McLeod SD DeBacker CM Viana MA . Differential care of corneal ulcers in the community based on apparent severity. Ophthalmology. 1996;103:479–484. [CrossRef] [PubMed]
Joseph J Murthy S Garg P Sharma S . Use of different stains for microscopic evaluation of corneal scrapings for diagnosis of microsporidial keratitis. J Clin Microbiol. 2006;44:583–585. [CrossRef] [PubMed]
Sawada Y Yuan C Huang AJW . Impression cytology in the diagnosis of Acanthamoeba keratitis with surface involvement. Am J Ophthalmol. 2004;137:323–328. [CrossRef] [PubMed]
Harvey RA Champe PC Fisher BD . Lippincott's Illustrated Reviews: Microbiology. 2nd ed. Philadelphia: Lippincott Williams & Wilkins; 2007.
Chen MT Wang HZ Chen CW . Evaluation of the diagnostic value of limulus amoebocyte lysate (LAL) test: bacterial culture and tissue smear in infectious keratitis. Taiwan J Ophthalmol. 1990;29:100–115.
Sharma S Kunimoto DY Gopinathan U Athmanathan S Garg P Rao GN . Evaluation of corneal scraping smear examination methods in the diagnosis of bacterial and fungal keratitis: a survey of eight years of laboratory experience. Cornea. 2002 21:643–647. [CrossRef] [PubMed]
Kanavi MR Javadi M Yazdani S Mirdehghanm S . Sensitivity and specificity of confocal scan in the diagnosis of infectious keratitis. Cornea. 2007;26:782–786. [CrossRef] [PubMed]
Tu EY Joslin CE Sugar J Booton GC Shoff ME Fuerst PA . The relative value of confocal microscopy and superficial corneal scrapings in the diagnosis of Acanthamoeba keratitis. Cornea. 2008;27:764–772. [CrossRef] [PubMed]
Parmar DN Awwad ST Petroll WM Bowman RW McCulley JP Cavanagh HD . Tandem scanning confocal corneal microscopy in the diagnosis of suspected Acanthamoeba keratitis. Ophthalmology. 2006;113:538–547. [CrossRef] [PubMed]
Kaufman SC Musch DC Belin MW . Confocal microscopy: a report by the American Academy of Ophthalmology. Ophthalmology. 2004;111:396–406. [CrossRef] [PubMed]
Byrne KA Burd EM Tabbara KF Hyndiuk RA . Diagnostic Microbiology and Cytology of the Eye. 1st ed. Boston: Butterworth-Heinemann; 1995.
Vengayil S Panda A Satpathy G . Polymerase chain reaction-guided diagnosis of mycotic keratitis: a prospective evaluation of its efficacy and limitations. Invest Ophthalmol Vis Sci. 2009;50:152–156. [CrossRef] [PubMed]
Kim E Chidambaram JD Srinivasan M . Prospective comparison of microbial culture and polymerase chain reaction in the diagnosis of corneal ulcer. Am J Ophthalmol. 2008;146:714–723. [CrossRef] [PubMed]
Fukuda M Deai T Higaki S Hayashi K Shimomura Y . Presence of a large amount of herpes simplex virus genome in tear fluid of herpetic stromal keratitis and persistent epithelial defect patients. Semin Ophthalmol. 2008;23:217–220. [CrossRef] [PubMed]
Kaji Y Hiraoka T Oshika T . Increased level of (1,3)-β-D-glucan in tear fluid of mycotic keratitis. Graefes Arch Clin Exp Ophthalmol. 2009;247:989–992. [CrossRef] [PubMed]
Ananthi S Chitra T Bini R Prajna NV Lalitha P Dharmalingam K . Comparative analysis of the tear protein profile in mycotic keratitis patients. Mol Vis. 2008;14.
Tei M Nishida K Kinoshita S . Polymerase chain reaction detection of herpes simplex virus in tear fluid from atypical herpetic epithelial keratitis after penetrating keratoplasty. Am J Ophthalmol. 1996;122:732–735. [CrossRef] [PubMed]
Prause JU . Serum albumin, serum antiproteases and polymorphonuclear leucocyte neutral collagenolytic protease in the tear fluid of patients with corneal ulcers. Acta Ophthalmol. 1983;61:272–282. [CrossRef]
Tervo T Salonen EM Vahen A . Elevation of tear fluid plasmin in corneal disease. Acta Ophthalmol. 1988;66:393–399. [CrossRef]
Fukuda M Mishima H Otori T . Detection of interleukin-1 beta in the tear fluid of patients with corneal disease with or without conjunctival involvement. Jpn J Ophthalmol. 1997;41:63–66. [CrossRef] [PubMed]
Gebhardt BM Varnell ED Kaufman HE . Acetylsalicylic acid reduces viral shedding induced by thermal stress. Curr Eye Res. 2004;29:119–125. [CrossRef] [PubMed]
Pramod NP Dhevahi E Sudhamathi K Ananda Kannan K Thyagarajan SP . Tear secretory IgA: evaluation of usefulness as a diagnostic marker in herpetic keratitis. Ocul Immunol Inflamm. 1999;7:61–67. [CrossRef] [PubMed]
McNamara NA Andika R Kwong M Sack RA Fleiszig SM . Interaction of Pseudomonas aeruginosa with human tear fluid components. Curr Eye Res. 2005;30:517–525. [CrossRef] [PubMed]
Cao Z Saravanan C Goldstein MH . Effect of human tears on Acanthamoeba-induced cytopathic effect. Arch Ophthalmol. 2008;126:348–352. [CrossRef] [PubMed]
Lehmann OJ Green SM Morlet N . Polymerase chain reaction analysis of corneal epithelial and tear samples in the diagnosis of Acanthamoeba keratitis. Invest Ophthalmol Vis Sci. 1998;39:1261–1265. [PubMed]
Thakur A Willcox MD . Cytokine and lipid inflammatory mediator profile of human tears during contact lens associated inflammatory diseases. Exp Eye Res. 1998;67:9–19. [CrossRef] [PubMed]
Eylan E Ronen D Romano A Smetana O . Lysozyme tear level in patients with herpes simplex virus eye infection. Invest Ophthalmol Vis Sci. 1977;16:850–853. [PubMed]
Raman CV Kirishnan KS . A new type of secondary radiation. Nature. 1928;121:105.
Lin CC Kuo MT Chang HC . Raman spectroscopy - a novel tool for noninvasive ocular surface fluid analysis. J Med Biol Eng. 2010;30:343–354. [CrossRef]
Barhoumi A Zhang D Tam F Halas NJ . Surface-enhanced Raman spectroscopy of DNA. J Am Chem Soc. 2008;130:5523–5529. [CrossRef] [PubMed]
Wei F Zhang D Halas NJ Hartgerink JD . Aromatic amino acids providing characteristic motifs in the Raman and SERS spectroscopy of peptides. J Phys Chem B. 2008;112:9158–9164. [CrossRef] [PubMed]
Lorincz A Haddad D Naik R . Raman spectroscopy for neoplastic tissue differentiation: a pilot study. J Pediatr Surg. 2004;39:953–956. [CrossRef] [PubMed]
Taleb A Diamond J McGarvey JJ Beattie JR Toland C Hamilton PW . Raman microscopy for the chemometric analysis of tumor cells. J Phys Chem B. 2006;110:19625–19631. [CrossRef] [PubMed]
Jess PR Smith DD Mazilu M Dholakia K Riches AC Herrington CS . Early detection of cervical neoplasia by Raman spectroscopy. Int J Cancer. 2007;121:2723–2728. [CrossRef] [PubMed]
Bauer NJ Motamedi M Hendrikse F Wicksted JP . Remote temperature monitoring in ocular tissue using confocal Raman spectroscopy. J Biomed Opt. 2005;10:031109.
Shen A Ye Y Wang X Chen C Zhang H Hu J . Raman scattering properties of human pterygium tissue. J Biomed Opt. 2005;10:024036.
Neelam K O'Gorman N Nolan J . Measurement of macular pigment: Raman spectroscopy versus heterochromatic flicker photometry. Invest Ophthalmol Vis Sci. 2005;46:1023–1032. [CrossRef] [PubMed]
Bernstein PS Zhao DY Wintch SW Ermakov IV McClane RW Gellermann W . Resonance Raman measurement of macular carotenoids in normal subjects and in age-related macular degeneration patients. Ophthalmology. 2002;109:1780–1787. [CrossRef] [PubMed]
Kim BS Lee CC Christensen JE Huser TR Chan JW Tarantal AF . Growth, differentiation, and biochemical signatures of rhesus monkey mesenchymal stem cells. Stem Cells Dev. 2008;17:185–198. [CrossRef] [PubMed]
Lambert JL Pelletier CC Borchert M . Glucose determination in human aqueous humor with Raman spectroscopy. J Biomed Opt. 2005;10:031110.
Hogg RE Zlatkova MB Chakravarthy U Anderson RS . Investigation of the effect of simulated lens yellowing, transparency loss and refractive error on in vivo resonance Raman spectroscopy. Ophthalmic Physiol Opt. 2007;27:225–231. [CrossRef] [PubMed]
Alexander TA . Development of methodology based on commercialized SERS-active substrates for rapid discrimination of Poxviridae virions. Anal Chem. 2008;80:2817–2825. [CrossRef] [PubMed]
Maquelin K Kirschner C Choo-Smith LP . Prospective study of the performance of vibrational spectroscopies for rapid identification of bacterial and fungal pathogens recovered from blood cultures. J Clin Microbiol. 2003;41:324–329. [CrossRef] [PubMed]
Sharma KD Andersson LA Loehr TM Terner J Goff HM . Comparative spectral analysis of mammalian, fungal, and bacterial catalases: resonance Raman evidence for iron-tyrosinate coordination. J Biol Chem. 1989;264:12772–12779. [PubMed]
López-Diez EC Winder CL Ashton L Currie F Goodacre R . Monitoring the mode of action of antibiotics using Raman spectroscopy: investigating subinhibitory effects of amikacin on Pseudomonas aeruginosa . Anal Chem. 2005;77:2901–2906. [CrossRef] [PubMed]
Smith E Dent G . Modern Raman Spectroscopy: A Practical Approach. Chichester: John Wiley & Sons; 2005.
Sengupta A Mujacic M Davis EJ . Detection of bacteria by surface-enhanced Raman spectroscopy. Anal Bioanal Chem. 2006;386:1379–1386. [CrossRef] [PubMed]
De Gelder J Willemse-Erix D Scholtes MJ . Monitoring poly(3-hydroxybutyrate) production in cupriavidus necator DSM 428 (H16) with Raman spectroscopy. Anal Chem. 2008;80:2155–2160. [CrossRef] [PubMed]
Zhang D Xie Y Mrozek MF Ortiz C Davisson VJ Ben-Amotz D . Raman detection of proteomic analytes. Anal Chem. 2003;75:5703–5709. [CrossRef] [PubMed]
Filik J Stone N . Drop coating deposition Raman spectroscopy of protein mixtures. Analyst. 2007;132:544–550. [CrossRef] [PubMed]
Filik J Stone N . Analysis of human tear fluid by Raman spectroscopy. Anal Chim Acta. 2008;616:177–184. [CrossRef] [PubMed]
Jolliffe IT . Principal Component Analysis. 2nd ed. New York: Springer; 2002.
Abdi H Williams LJ . Principal component analysis. WIREs: Comput Stat. 2010;2:433–459. [CrossRef]
Guo Q Wu W Massart DL Boucon C de Jong S . Feature selection in principal component analysis of analytical data. Chemometr Intell Lab. 2002;61:123–132. [CrossRef]
Hu J Si J Olson BP He J . Principle component feature detector for motor cortical control. Conf Proc IEEE Eng Med Biol Soc. 2004:4021–4024.
De Gelder J De Gussem K Vandenabeele P Vancanneyt M De Vos P Moens L . Methods for extracting biochemical information from bacterial Raman spectra: focus on a group of structurally similar biomolecules—fatty acids. Anal Chim Acta. 2007;603:167–175. [CrossRef] [PubMed]
Kaye S Tuft S Neal T . Bacterial susceptibility to topical antimicrobials and clinical outcome in bacterial keratitis. Invest Ophthalmol Vis Sci. 2010;51:362–368. [CrossRef] [PubMed]
Deegan RD Bakajin O Dupont TF Huber G Nagel SR Witten TA . Capillary flow as the cause for ring stains from dried liquid drops. Nature. 1997;389:827–829. [CrossRef]
Kuo MT Lin CC Liu HY Chang HC . Tear analytical model based on Raman microspectroscopy for investigation of infectious diseases of the ocular surface. Invest Ophthalmol Vis Sci. 2011;52:4942–4950. [CrossRef] [PubMed]
Rossi EE Silveira LJr Barbosa Pinheiro AL . Raman spectroscopy for differential diagnosis of endophthalmitis and uveitis in rabbit iris in vitro. Exp Eye Res. 2010;91:362–368. [CrossRef] [PubMed]
Lasch P Kneipp J . Biomedical Vibrational Spectroscopy: Hoboken, NJ: Wiley-Interscience; 2008.
Filik J Stone N . Investigation into the protein composition of human tear fluid using centrifugal filters and drop coating deposition Raman spectroscopy. J Raman Spectrosc. 2009;40:218–224. [CrossRef]
Willemse-Erix DF Scholtes-Timmerman MJ Jachtenberg JW . Optical fingerprinting in bacterial epidemiology: raman spectroscopy as a real-time typing method. J Clin Microbiol. 2009;47:652–659. [CrossRef] [PubMed]
Reyes-Goddard JM Barr H Stone N . Surface enhanced Raman scattering of herpes simplex virus in tear film. Photodiagnosis Photodyn Ther. 2008;5:42–49. [CrossRef] [PubMed]
Jayamanne DG Dayan M Jenkins D Porter R . The role of staphylococcal superantigens in the pathogenesis of marginal keratitis. Eye. 1997;11:618–621. [CrossRef] [PubMed]
Özcura F . Successful treatment of staphylococcus-associated marginal keratitis with topical cyclosporine. Graefes Arch Clin Exp Ophthalmol. 2010;248:1049–1050. [CrossRef] [PubMed]
Reading NC Sperandio V . Quorum sensing: the many languages of bacteria. FEMS Microbiol Lett. 2006;254:1–11. [CrossRef] [PubMed]
Tam C Mun JJ Evans DJ Fleiszig SM . The impact of inoculation parameters on the pathogenesis of contact lens-related infectious keratitis. Invest Ophthalmol Vis Sci. 2010;51:3100–3106. [CrossRef] [PubMed]
Hogan DA . Talking to themselves: autoregulation and quorum sensing in fungi. Eukaryot Cell. 2006;5:613–619. [CrossRef] [PubMed]
Willcox MD . Pseudomonas aeruginosa infection and inflammation during contact lens wear: a review. Optom Vis Sci. 2007;84:273–278. [CrossRef] [PubMed]
Graham JE Moore JE Jiru X . Ocular pathogen or commensal: a PCR-based study of surface bacterial flora in normal and dry eyes. Invest Ophthalmol Vis Sci. 2007;48:5616–5623. [CrossRef] [PubMed]
Figure 1.
 
Mean Raman spectra of tear samples with 95% confidence intervals of different tear groups, including healthy subjects (300 spectra of 30 subjects), noninfectious keratitis (260 spectra of 26 patients), and infectious keratitis (260 spectra of 26 patients) (a). Difference in the Raman spectra of tear samples with 95% confidence intervals of different tear groups (b).
Figure 1.
 
Mean Raman spectra of tear samples with 95% confidence intervals of different tear groups, including healthy subjects (300 spectra of 30 subjects), noninfectious keratitis (260 spectra of 26 patients), and infectious keratitis (260 spectra of 26 patients) (a). Difference in the Raman spectra of tear samples with 95% confidence intervals of different tear groups (b).
Figure 2.
 
PC transformation of Raman spectra of tear samples is based on the Norm group. The principal vectors of Norm serve as a set of transfer functions to extract global tear features for different tear groups (a). PC1 to PC6 had a cumulated relevance of approximately 80% for the Norm group, and therefore, serve as global features of Raman spectra of tear samples (b).
Figure 2.
 
PC transformation of Raman spectra of tear samples is based on the Norm group. The principal vectors of Norm serve as a set of transfer functions to extract global tear features for different tear groups (a). PC1 to PC6 had a cumulated relevance of approximately 80% for the Norm group, and therefore, serve as global features of Raman spectra of tear samples (b).
Figure 3.
 
Comparison of the scores of the Norm-based PCs between IUK (red block) and NIUK (black block). The principal scores were acquired on the basis of the transformation system of normal teardrops. Each box is constructed by five parameters, including the median (Q2), lower, and upper quartiles (Q1, Q3), and lowest and highest data (Q1 − 1.5 × [Q3 − Q1], Q3 + 1.5 × [Q3 − Q1]). One-tailed t-test, *P < 0.05. + Power > 0.8.
Figure 3.
 
Comparison of the scores of the Norm-based PCs between IUK (red block) and NIUK (black block). The principal scores were acquired on the basis of the transformation system of normal teardrops. Each box is constructed by five parameters, including the median (Q2), lower, and upper quartiles (Q1, Q3), and lowest and highest data (Q1 − 1.5 × [Q3 − Q1], Q3 + 1.5 × [Q3 − Q1]). One-tailed t-test, *P < 0.05. + Power > 0.8.
Figure 4.
 
Comparison of 95% confidence intervals of the significant scores between the IUK and Norm groups. Exclusive comparisons among different microbial subgroups of IUK for the scores of PC1 (a), PC2 (c), and PC4 (e). Inclusive comparisons among overlapping bacterial subgroups for the scores of PC1 (b), PC2 (d), and PC4 (f).
Figure 4.
 
Comparison of 95% confidence intervals of the significant scores between the IUK and Norm groups. Exclusive comparisons among different microbial subgroups of IUK for the scores of PC1 (a), PC2 (c), and PC4 (e). Inclusive comparisons among overlapping bacterial subgroups for the scores of PC1 (b), PC2 (d), and PC4 (f).
Table 1.
 
Clinical Data of the Three Participant Groups
Table 1.
 
Clinical Data of the Three Participant Groups
Clinical Features Norm Group IUK Group NIUK Group
Case number 30 26 26
Age, y 42.7 ± 13.5 46.5 ± 17.8 46.1 ± 21.3
Sex, male/female 15/15 9/17 12/14
Laterality, OD/OS 16/14 11/15 12/14
Risk factors for keratitis
    Contact lens wear 9 7
    Corneal trauma 10 10
    Systemic disease 2 2
    Ocular surface and corneal disease 2 6
    Previous ulcerative keratitis 3 1
Topical antimicrobials before visit
    No 6 5
    Yes, ≤3 days 8 8
    Yes, >3 days 12 13
Boundary pattern of ulcer
    Near elliptical shape 19 10
    Nonelliptical shape 7 16
Area of ulcer, mm2 10.2 ± 9.3 6.5 ± 10.8
Depth of ulcer (anterior one-third/posterior two-thirds) 17/9 21/5
Location of ulcer (central/paracentral/peripheral) 12/11/3 8/9/9
Figure sf01, PDF
Table st1, PDF
Table st2, PDF
×
×

This PDF is available to Subscribers Only

Sign in or purchase a subscription to access this content. ×

You must be signed into an individual account to use this feature.

×