Usc of Optica l Imaging as II Diagnostic Tool for Idcntilication of Pathogenic Markers associated with 1-llIllIan Sera

Abstract

In this research thesis the potential of optical diagnostic techniques i.e. Raman and ATR-FTIR spectroscopy were evaluated for identification of specific pathogenic markers for discrimination between healthy and infected sera. One bacterial (S. typhi) and two viral (DENV and HCV) infected human blood sera were studied and classified as compared to healthy sera in the same age and gender. Inclusion of a bacterial infection along with two viral infections ruled out biochemical signatures arising due to fever only making it easier to distinguish infection specific signatures. The major aim of the study was to look for pathogen specific markers that can be used to develop an alternative diagnostic tool, which is cost-effective and provide efficient diagnosis of the three major endemic pathogenic infections prevalent in Pakistan. Raman is a non-invasive technique that requires small amount of sample with minimal preparation and its spectral acquisition is not hampered by the water molecules in biological fluid. Sera samples were collected from infected individuals for S. typhi, DENV and HCV. These individuals were screened using typhidot (dot ELISA) for S. typhi and PCR for viral infection to confirm the presence of respective infections. Sera from healthy volunteers were also collected and all the sampling was carried out with informed consent. For spectral acquisition, 20 µL of serum was placed in a crevice on the surface of the aluminum block. Spectra were recorded using Peak Seeker PRO-785 coupled with microscope RSM-785, Agiltron, USA Raman system. Spectra obtained have a spectral resolution of 10 cm−1 and excitation wavelength 785 nm. Twelve spectra were obtained from every sample using laser light of 150 mW for 10-s integration time. Raw spectra were pre-processed to remove the noise. For this purpose, smoothing, baseline correction, and vector normalization were performed in MATLAB (Mathworks, release 2009; Natick, MA, USA). This study for the first-time reported infection specific Raman signatures from S. typhi infected sera as seven unique bands at 630, 1099, 1144, 1343, 1692, 1711 and 1746 cm-1 containing significant spectral variations. At these bands PCA classified the infected and healthy samples with 90 % accuracy. To unveil the efficacy of Raman spectroscopy for DENV diagnosis the spectral signatures identified in this study were compared with previously reported spectral signatures associated with this virus and PCA-LDA model was developed. Based on the five peaks consistently associated with DENV at 850, 1333, 1450, 1580 and 1672 cm-1 PCA-LDA model provided 100 % classification accuracy for discrimination of DENV infection from healthy sera. The current work also presents clear differences between Raman spectra of HCV infected and healthy sera samples. Using ANOVA twelve unique Raman bands at 676, 825, 853, 936, 1029, 1105, 1155, 1180, 1305, 1330,1620, 1654 and 1757 cm−1 associated with only HCV infected sera which have not been reported in earlier studies were identified. We used PCA in conjunction with LDA, on these set of signatures using ten-fold jackknife cross-validation that yielded 98 % sensitivity and specificity. ATR-FTIR spectroscopy makes use of crystal that interacts with samples under examination. A beam of infra-red (IR) radiations passes through the crystal, and then the xii beam is internally reflected without being refracted. Due to this reflection, a transitory wave is generated that penetrates the sample, giving rise to a characteristic spectrum. Freeze dried sera were used for the spectral acquisition as presence of moisture hampers spectra recorded from biological fluid. Freeze-dried sera samples were used for spectral acquisition using BRUKER FTIR model Alpha II Platinum FTIR. Background noise was subtracted and small amount of freeze-dried sample was placed onto the ATR crystal. Spectra were recorded in IR region i.e. 400–4000 cm−1 , resolution 4 cm−1 with 32 scans per sample. After spectral acquisition data was processed using Unscrambler X version 10.5. The spectra were vector normalized using Standard Normal Variate (SNV) transformation in order to minimize the scattering effects and background noise and enhance inter-sample differences. For a more enhanced discrimination between the two classes multivariate data classification algorisms were applied. This study is the first report on use of freeze-dried sera for recording ATR-FTIR spectral signatures from S. typhi, DENV and HCV infected sera. For classification of S. typhi infected sera PLSR model predicted infection excellently with R2 value of 0.9757 and HCA dendogram showed 100 % sensitivity and 96 % specificity. Investigation of ATR-FTIR spectral markers associated with DENV using freeze-dried sera samples in this study reports 100 % classification accuracy using PCA and sensitivity and specificity of 89 % and 95 % respectively using PCA-LDA model. PLSR prediction showed excellent differentiation into two classes based on biochemical changes as depicted by the value of R2 = 0.9980. For identification of ATR-FTIR specific signatures associated with HCV infected sera a PCA-LDA classification model was build. The average spectra of healthy and HCV infected sera samples are discriminated in these studies at 3500- 2700 cm-1 . Based on obtained results and previous literature it is proposed that Raman and ATR-FTIR hold the potential to not only identify but also to differentiate between different pathological conditions. Moreover, the interpretation of outcomes is independent of a pathologist’s observation, making these fast detection tools. However, more clinical trials with larger sample size are needed to accept the reported set of spectroscopic signatures for clinical diagnostics. Unification of sampling techniques and processing methods will be advantageous and might lead to immense improvement in clinical implementation of these methods as next generation diagnostic tools