REAL-TIME BIOBURDEN AND DIRECT ENDOTOXIN SCREENING Cross-Reference to Related Applications This application claims the benefit of U.S. Provisional Application No.63/310,271, entitled “REAL-TIME BIOBURDEN AND DIRECT ENDOTOXIN SCREENING,” filed February 15, 2022, the entirety of which is incorporated herein by reference. Statement Regarding Federally Sponsored Research This invention was made with Government support under Award No.1632420 awarded by the National Science Foundation. The Government has certain rights in this invention. Field of the Invention This invention relates to a system and method for rapid real-time bacterial and/or endotoxin detection during routine sampling of in manufacturing operations, and in particular to rapid detection of microbial organisms associated with bacteria and/or endotoxins harvested during routine sampling in manufacturing operations, not requiring the growth and reproduction phase of the microbial organism. Background The quantitative assessment of microbial presence in pharmaceutical, medical, food, water, drug, environmental, or biological manufacturing processes traditionally has involved a growth-based approach which has been harmonized by the U.S. pharmacopeia. This process is labor intensive and time consuming, and is limited by the growth medium used and incubation conditions which typically involves up to 14 days and at times 21 days depending on the microbial strain being evaluated. Thus, this growth-based approach is required among others for lot release testing which results in long inventory holding periods and delays in manufacturing and distribution. Detection of microbial presence may be conducted using infrared spectroscopy. Infrared spectroscopy allows for the assessment of the basic biochemical components of microorganisms by allowing the identification of lipids, proteins, DNA, RNA and carbohydrates. The characterization of bacteria has been carried out since the 1950’s, and the barrier to widespread adoption was the time required to acquire the spectral data. Improved spectral acquisition times were accomplished by using the combined approach of Fourier   transform infrared (FT-IR) spectroscopy and chemometrics for the classification of bacteria, requiring the creation of a spectral library or database. In order for these approaches to work, the microorganisms need to be grown in media for selection, analysis and comparison. More recently, the implementation of mass spectroscopy (MS) and FT-IR have been proposed for the classification of the bacteria, which also requires the growth and reproduction of the microorganism under controlled conditions to optimize for the species. Still further improvement is needed for the screening of microbial presence in a given sample to ensure rapid, near real-time assessment. Moreover, the concentration and homogeneity of the sample is often a cause for concern when implementing rapid methods of detection for screening. The presence of pathogens such as bacteria in the manufacturing setting poses a risk to the quality of the product, such as a drug substance or drug product, being produced. The biopharma industry employs different living organisms selected to produce the quality attributes to ensure efficacy of a therapeutic protein for the treatment of a specific disease. The production of these drugs must be free of bacterial contamination. Also, the screening of water, food, and other products destined for human consumption or use requires that such products are free of bacterial contamination. For example, there are five pathogens – Staphylococcus aureus, Escherichia coli, Streptococcus pneumoniae, Klebsiella pneumoniae, and Pseudomonas aeruginosa – that are responsible for a large number of deaths when consumed by humans. The prevalence of different pathogens and bacterial strains may vary by geographic region, and in the biopharma industry the source of bacterial contamination varies from water, water systems, incoming raw materials, lack of adequate/effective process controls, and human activity. Two species of bacteria which cause contamination concerns are known as: (1) Gram positive with a peptidoglycan outer cell wall and (2) Gram negative with an outer cell membrane that contains lipopolysaccharides (LPS) that is shed upon death of the microorganism and, both of which are known to induce an unwanted immune response that can be lethal. With respect to Gram negative, once the outer shell membrane containing the LPS is shed, the LPS is referred to as endotoxin. Biofilms are of particular concern with water systems, since Gram negative bacteria constitute the majority of the bacterial populations found in aqueous environments. Endotoxin detection has been performed almost exclusively by employing an enzymatic assay known as Limulus Amebocyte Lysate (LAL) assay. The enzyme is harvested from the blue blood of the horseshoe crab. Recently, the use of a recombinant alternative has also been introduced for the detection of the endotoxin in solution. These   enzyme-based assays are known to have inhibitory substances and false positive results. To date, the alternative is to repeat the assay to eliminate conflicting results and at times recruit to a qualitative test. Therefore, the need for a direct and rapid method of detection exists. The Food and Drug Administration (FDA) as well as other relevant regulatory agencies have provided guidelines for the detection of bioburden and endotoxin in the biopharma setting. The guidelines establish the screening for microbial presence. For the biopharmaceutical industry, manufacturing operations (upstream and downstream processing) screening of complex samples is performed following the current compendial methods. For example, during upstream processing cell cultures are used for the production of the desired therapeutic protein. The bacterial contamination may also occur during downstream processing as well. Bacterial contamination may occur during the drug development cycle and has been the cause of not allowing for lot release or withdrawal. Thus, a need exists for a rapid, reliable method for detection of bacterial or endotoxin contamination of samples, and for monitoring the growth of any bacterial cells. Summary of the Invention The system and method described herein provide a rapid method of real-time bacterial and/or endotoxin detection of the microbial organisms harvested during routine sampling in manufacturing operations, not requiring the growth and reproduction phase of the microbial organism. Bioburden assessment follows compendial methods for raw materials, nonsterile drug substances and drug products including release testing. The time savings provides a competitive advantage to the Biopharma industry for fast-turn-around times of the screening process accelerating time to market. Rapid method of detection is assured only when a direct real-time screening method can be employed, as described with respect to the system and method. The system and method use a vibrational spectroscopy apparatus, such as a quantum cascade laser (QCL) transmission microscope, Raman amplifier, FT-IR, or light detection and ranging (LIDAR) system, with a slide cell array and dedicated software that provide direct screening of samples for bacterial and endotoxin presence. The slide cell array may be comprised of a polymer, such as polyethylene, calcium fluoride, barium fluoride, or other polymer that is transparent within most of the spectral region of interest. This direct screening may be conducted in real-time, including when a QCL microscope is used to obtain images of the slide array. The samples to be screened, along with potentially one or more controls, are provided within the cells of the slide cell array, and the QCL is used to acquire images of the   slide cell array. The system may be implemented as a platform, with data from the QCL provided to the dedicated software for analysis of the samples. The QCLs provide real-time hyperspectral (HS) image acquisition with enhanced signal to noise ratio. The QCL contains a light source, and the intensity of the light source may be comparable to the synchrotron and is non-destructive. Each hyperspectral image obtained is comprised of 223,000 spectra. As mentioned, the hyperspectral images may also be acquired by other vibrational spectroscopy apparatus, such as a Raman amplifier or LIDAR. The slide cell array and slide cell holder accessory ensure a controlled thermal environment for up to 23 samples screened at a fixed pathlength between 3-5 ^m. The slide cell holder accessory may be, for example, a controllable heated chamber configured to receive the slide cell. The hyperspectral images are evaluated and both two-dimensional correlation spectroscopy (2D-COS), and/or two- dimensional co-distribution spectroscopy (2D-CDS) are employed to correlate the presence of both bacteria and endotoxin. These correlations ensure the determination of the bacterial and endotoxin presence. The analysis is based on the principle of differences in growth rates between the organism being evaluated and that of the bacterial contaminant allowing for the determination of cross peaks associated with different biochemical components be designated directly to the bacteria within the sample. No a priori knowledge nor the need for databases is required. The platform technology can be implemented in manufacturing operations, such as biopharma operations, in both upstream as well as downstream processes to provide real-time or near real-time bacterial and direct endotoxin screening, thereby providing comprehensive regulatory compliance. In addition, the system and method may screen live sample, and then spike another live sample with a known amount of the endotoxin. The system may then then text both the live sample and the spiked sample, and if the live sample correlates to the results for the spiked sample, then indicate a positive result is indicated for the live sample. This approach can also be added on as a check to the method recited above where the comparison is between live and dead samples to confirm that a positive result is actually positive. Bacterial Detection For example, a method for detecting a microorganism in a sample may comprise: a) providing the sample in a slide containing at least one sample well; b) acquiring at least one first spectral image of the sample using a quantum cascade laser microscope, wherein the slide provides a fixed path length of between 3-5µm;   c) identifying and selecting, in at least one of the acquired first spectral images, a region of interest; d) obtaining spectral data for the region of interest; e) applying a baseline correction to the spectral data for the region of interest; f) determining peak intensities of the baseline corrected spectral data at wavenumbers corresponding to vibrational modes for: i) amide I and amide II; and ii) at least two of carbohydrates (C-O and COO-), and membrane associated vibrational modes (PO2 and CH2); g) determining an amide II to amide I peak intensity ratio, and determining whether this amide II to amide I peak intensity ratio is within a predetermined threshold; h) when the determined amide II to amide I peak intensity ratio is within the predetermined threshold, determining a peak intensity ratio for at least one of the determined peak intensities determined in step f), ii) relative to at least one other determined peak intensity in step f), ii), i) determining whether one or more of the determined peak intensity ratios in step h) increases relative to a suitable control; and j) indicating the presence of a microorganism when one or more of the peak intensity ratios in step h) increases relative to the suitable control. The method may further comprise: acquiring at least one second spectral image using a quantum cascade laser microscope, wherein the at least one second spectral image is acquired for the sample at a different concentration or after the application of heat to the sample; obtaining spectral data for the at least one second spectral image; and applying two-dimensional co-distribution analysis to the obtained spectral data for the at least one second spectral image to generate a co-distribution plot for the sample; determining whether the co-distribution plot includes at least one cross peak associated with the presence of the microorganism in the sample; and confirming the indication of the presence of the microorganism in the sample when the at least one cross peak is included in the co-distribution plot. The method may further comprise analyzing the co-distribution plot, and determining the sample is homogeneous when peaks in the co-distribution plot are at expected positions. The method may further comprise applying two-dimensional correlation (2D-COS) analysis to generate a synchronous correlation plot and an asynchronous   correlation plot; and identifying in the synchronous correlation plot cross peaks that correlate with the presence of the microorganism. In the method, the region of interest may be the full field of the at least one acquired first spectral image, or the region of interest may be a portion of the at least one acquired first spectral image. In the method, the wavenumber for carbohydrates is 1084 +-2 cm ^-1 , the wavenumber for PO2 is 1156 +-2 cm ^-1 and/or 1236 +- 2 cm ^-1 , the wavenumber for CH2 is 1454 +-2 cm ^-1 , and the wavenumber for C=O, COO- is 1404 +-2 cm ^-1 . The predetermined threshold for the ratio of amide II to amide I peak intensity may be 0.4 to 0.6. Further, acquiring at least one first spectral image of the sample using a quantum cascade laser microscope may include acquiring at least one first hyperspectral image of the sample, and the second spectral images may also be hyperspectral images. The suitable control for the method may be water. Endotoxin Detection: A method for detecting an endotoxin in a sample comprises: a) providing a first amount of the sample in at least one sample well and providing a second amount of the sample that has been treated to kill any microbes that are present; b) acquiring at a first spectral image of the first sample amount and a second spectral image of the second sample amount using a quantum cascade laser microscope, wherein the slide provides a fixed path length of between 3-5µm; c) identifying and selecting, in each of the first and second spectral images, a region of interest; d) obtaining first spectral data for the region of interest in the first spectral image and second spectral data for the region of interest in the second spectral image; e) applying a baseline correction to the first and second spectral data; f) determining peak intensities of the baseline corrected first and second spectral data at wavenumbers corresponding vibrational modes for: i) amide I and amide II; and ii) at least two of carbohydrates, PO2, CH2, C-O and COO-; g) determining an amide II to amide I peak intensity ratio for each of the first and second spectral data, and determining whether the amide II to amide I peak intensity ratios are within a predetermined threshold;   h) when the determined amide II to amide I peak intensity ratios are within the predetermined threshold: i) determining a peak intensity ratio for at least one of the determined peak intensities determined in step f), ii) relative to at least one other determined peak intensity in step f), ii) for the first spectral data; and ii) determining a peak intensity ratio for at least one of the determined peak intensities determined in step f), ii) relative to at least one other determined peak intensity in step f), ii) for the second spectral data; i) comparing the determined peak intensity ratios in step h), i) to the corresponding peak intensity ratios in step h), ii) and generating test ratios based on the comparisons; i) indicating the presence of an endotoxin when one or more of the test ratios is within a predetermined test ratio threshold. The pre-determined test ratio threshold may be less than or equal to about 0.6 or greater than or equal to about 1.4. The method may further comprise: acquiring at least one third spectral image using a quantum cascade laser microscope, wherein the at least one third spectral image is acquired for the sample at a different concentration or after the application of heat to the sample; obtaining spectral data for the at least one third spectral image; and applying two-dimensional co-distribution analysis to the obtained spectral data for the at least one third spectral image to generate a co-distribution plot for the sample; determining whether the co-distribution plot includes at least one cross peak associated with the presence of the microorganism in the sample; and confirming the indication of the presence of endotoxin in the sample when the at least one cross peak is included in the co-distribution plot. The method may further comprise analyzing the co-distribution plot, and determining the sample is homogeneous when peaks in the co-distribution plot are at expected positions. In the method, the region of interest may be the full field of the at least one acquired first spectral images, or the region of interest may be a portion of the at least one acquired first spectral images. In the method, the wavenumber for carbohydrates is 1084 +-2 cm ^-1 , 1115 +-2 cm-1, 1038 +-2 cm ^-1 , the wavenumber for PO2 is 1156 +-2 cm ^-1 and/or 1236 +- 2 cm ^-1 , the wavenumber for CH ₂ is 1454 +-2 cm ^-1 , and the wavenumber for C=O, COO- is 1404 +-2 cm- ¹ .   The pre-determined threshold for the ratio of amide II to amide I peak intensity may be greater than about 0.35. The method may further comprise: acquiring at least one third spectral image using a quantum cascade laser microscope, wherein the at least one third spectral image is acquired for the sample at a different concentration or after the application of heat to the sample; obtaining spectral data for the at least one third spectral image; and applying two-dimensional correlation analysis to the obtained spectral data for the at least one third spectral image to generate synchronous and asynchronous correlation plots for the sample; determining whether the synchronous and asynchronous correlation plots include at least two cross peaks associated with the presence of the microorganism in the sample; and confirming the indication of the presence of endotoxin in the sample when the at least two cross peaks are included in the synchronous and asynchronous correlation plots. The first, second, and third spectral images may be hyperspectral images. In the method, treating the sample to kill any microbes that are present comprises heat treating, ultraviolet light, or chemical treating. 2T2D Endotoxin Detection A method for detecting a microorganism in a sample comprises: a) providing the sample in a slide containing at least one sample well; b) acquiring at least one first spectral image of the sample using a quantum cascade laser microscope; c) identifying and selecting, in at least one of the acquired first spectral images, a region of interest; d) obtaining first spectral data for the region of interest, and applying a baseline correction to the first spectral data for the region of interest; e) providing second spectral data comprising reference spectral data of an uncontaminated sample; f) applying a two-trace two-dimensional (2T2D) correlation to the baseline corrected first spectral data and second spectral data to generate a synchronous spectrum Φ nd asynchronous spectrum Ψ^ ^^ ^^ ^ g) generating a synchronous plot and an asynchronous plot from the synchronous and asynchronous spectrum; and   h) identifying cross peaks in the asynchronous plot, and associating positive cross peaks with a first component in the sample and associating negative cross peaks with a second component in the sample. The method may further comprise generating a first weighted difference spectrum for the baseline corrected first spectral data, generating a second weighted difference spectrum for the second spectral data, identifying key cross-peaks in the first weighted spectrum and second weighted spectrum, and determining components present in the sample based on the sign of the identified cross peaks. The positive cross peaks in the weighted spectra are determined to indicate the first component in the sample, and wherein negative cross peaks are determined to indicate the second component within the sample. The method may also include applying a 2T2D correlation coefficient , to the synchronous spectrum, and applying a disrelation coefficient to the asynchronous spectrum to scale the 2T2D correlation spectra, as given by: Brief Description of the Drawings FIG.1 is an illustration of a slide cell sample array and resultant images and data extracted from the sample by the system and method. FIG.2 is a flow diagram of processing steps for determining whether acceptance criteria are met in the bacteria and/or endotoxin screening process. FIG.3 is a flow diagram of processing steps for evaluating whether bacteria and/or endotoxin are present in a sample. FIGs.4A and 4B are an example illustration of a hyperspectral image (Fig.4A) along with an overlay of the QCL infrared microscope (QCL IRM) transmission spectra (Fig.4B). FIG.5 provides an example of the linearity and Orthogonality for optical density 600 nm using a quantum cascade laser microscope for three Gram negative bacterial species with key peaks identified. FIG.6 provides an example of the linearity and orthogonality for optical density 600 nm using a quantum cascade laser microscope for three Gram positive bacterial species, with key peaks identified.   FIGs.7A and 7B are an example of a 2D co-distribution asynchronous plot for E. coli within the spectral region of: 1780 – 1000 cm ^-1 (Fig.7A) and 1482 -– 1000 cm ^-1 (Fig.7B). FIGs.8A-8D are an example of 2D-COS plots for E. coli within the spectral region of (top row) 1780 – 1000 cm ^-1 and (bottom row) 1482 - 1000 cm ^-1 , which are used to investigate further the bacterial contamination within the sample, where Figs.8A, 8C are synchronous plots and Figs.8B, 8D are asynchronous plots. FIGs.9A and 9B are an example of a 2D co-distribution asynchronous plot for Pseudomonas aeruginosa within the spectral region of: 1780 – 1000 cm ^-1 (Fig.9A) and 1482 -– 1000 cm ^-1 (Fig.9B). FIGs.10A-10D are an example of 2D-COS plots for Pseudomonas aeruginosa within the spectral region of (top row) 1780 – 1000 cm ^-1 and (bottom row) 1482 - 1000 cm ^-1 , which are used are used to investigate further the bacterial contamination within the sample. Figs. 10A and 10C are synchronous plots and Figs.10B and 10D are asynchronous plots. FIG.11 is a schematic diagram of a typical growth curve for bacteria over time, illustrating periods of lag, growth, stationary development, and death. FIG.12A is an example of a growth curve for Pseudomonas aeruginosa, showing the signature peak absorbances (A.U.) over time for the 1085 cm ^-1 signature peak. FIG.12B is an example of a growth curve for Pseudomonas aeruginosa, showing the signature peak absorbances (A.U.) over time for the 1237 cm ^-1 signature peak. FIG.13A is an example of a growth curve for Pseudomonas aeruginosa, showing the signature peak absorbances (A.U.) over time for the 1404 cm ^-1 signature peak. FIG.13B is an example of a growth curve for Pseudomonas aeruginosa, showing the signature peak absorbances (A.U.) over time for the 1454 cm ^-1 signature peak. FIG.14 is an illustration of a slide sample array containing at least three rows of sample wells, in which triplicate samples can be taken at different points in time during a bacterial cell culture. FIG.15A and 15B show the 2T2D synchronous (Fig.15A) and asynchronous (Fig. 15B) correlation plots for a P. aeruginosa cell culture in the presence of endotoxin in the spectral region of 1500 – 1010 cm ^-1 . FIGs.16A and 16B show the weighted difference spectra for both components of a sample, with FIG.16A showing the weighted difference spectra for the endotoxin component and FIG.16B showing the weighted difference spectra for the P. aeruginosa cells. Characteristic main spectral peaks observed for each component are observed providing evidence of the same.   FIGs.17A and 17B show the co-distribution asynchronous plots of mammalian cell culture samples harvested as a function of time during a run for experimental growth (Fig. 17A) and experimentally simulated dying conditions (Fig.17B). FIG.18 is a schematic workflow of mammalian cell cultures being spiked with known amount of E. coli for evaluation using the methods described herein. FIG.19A-19D show average absorbances for mammalian culture cells spiked with E. coli, as absorbance vs. cells/mL, for key signature peaks of 1084 cm ^-1 , 1235 cm ^-1 , 1454 cm ^-1 , and 1404 cm ^-1 , respectively. FIG.19E shows how the culture cells were spiked with E. coli. FIGs.20A and 20B show 2D-CDS asynchronous plots for mammalian cells (Fig. 20A) and mammalian cells spiked with E. coli (Fig.20B) in the spectral region of 1485 – 1000 cm ^-1 . FIG.21A shows a 2D-COS synchronous plot for experimentally simulated dying mammalian cells for experimental data within the spectral region 1485 – 1000 cm ^-1 . FIG.21B shows a 2D-COS synchronous plot for experimentally simulated dying mammalian cells in the presence of increasing E. coli cells for experimental data within the spectral region 1485 – 1000 cm ^-1 . FIG.21C shows a 2D-COS asynchronous plot for experimentally simulated dying mammalian cells for experimental data within the spectral region 1485 – 1000 cm ^-1 . FIG.21D shows a 2D-COS asynchronous plot for experimentally simulated dying mammalian cells in the presence of increasing E. coli cells for experimental data within the spectral region 1485 – 1000 cm ^-1 . FIG.22A is an example of a hyperspectral image at 4.3 ^m spatial resolution for a mammalian cell culture in the presence of Gram negative bacteria control standard endotoxin (spiked). FIG.22B is an example of an enhanced QCL infrared microscopy spectrum for the hyperspectral image of Figure 22A. FIGs.23A and 23B show 2T2D correlation plots for mammalian cell culture in the presence of control standard endotoxin (spiked) mixture and the mammalian cell culture as reference in the spectral region of 1750 – 1010 cm ^-1 , with Figure 23A showing the synchronous plot and Figure 23B showing the asynchronous plot. FIG.24A shows weighted difference spectra for control standard endotoxin, illustrating the main spectral components associated with endotoxin including the amide of   the lipid carbonyl at 1726 cm ^-1 , the glucosamine component at 1620 cm ^-1 , the methylene in- plane bending 1454 cm ^-1 , carbohydrate stretching (C=O), COO-, phosphodiester asymmetric and symmetric stretches at 1237 and 1156 cm ^-1 , respectively; and carbohydrates vibrational modes at 1117, ~1080 and ~1060 cm ^-1 of the endotoxin. FIG.24B shows weighted difference spectra for a mammalian cell culture, illustrating the main spectral components, being the amide I, II and III bands at 1645, 1550 and ~1365 cm ^-1 , associated with mammalian cells. FIG.25 is an illustration of upstream processing and downstream processing in a manufacturing process, including indications of points along the processing steps where bacteria and endotoxin screening can be conducted. Detailed Description of the Drawings FIGs.1-3 provide a diagram of a sample cell array and images and spectral analysis associated with the system and method of bacteria and/or endotoxin detection, and flow diagrams of processing steps in the method. Samples and appropriate controls are placed in pre-defined wells within the slide cell array, such as that shown in FIG.1. Hyperspectral images are acquired at one or more defined temperatures. Initially, the amide II / amide I band ratio within the QCL IRM spectra are evaluated for H2O content to ensure reproducibility of results. The acceptance criteria threshold value for the amide II/amide I ratio is ^ 0.5. The accepted QCL IRM transmission spectra, that is those that met the acceptance criteria, are then evaluated for linearity and orthogonality thus allowing for comparability assessment of the spectral data acquired. A QCL IRM covariance spectra is then subject to two-dimensional co-distribution (2D-CDS) and two-dimensional correlation analysis (2D-COS). The correlations observed from the 2D-CDS asynchronous plot allow for Go or No-Go decision making for bacterial presence. Furthermore, the correlations involving key signature peaks are used to evaluate the complexity of the chemical composition within each the sample in the array. A typical hyperspectral image along with an overlay of the QCL IRM transmission spectra is shown in FIGs.4A and 4B, respectively. The QCL IRM spectra allow for the direct determination key signature peaks in the spectral region of interest, such as 1780 – 1000 cm ^-1 for both Gram negative and/or Gram positive bacteria. The method may reference the American Type Culture Collection (ATCC) certified reference material for a series of bacterial strains. The QCLM orthogonality with optical density (OD) 600 nm and linearity   results are summarized in FIGs.5 and 6, providing a linear relationship with the bacterial cell concentration (cells/mL). Furthermore, these signature peaks allow for the correlation of the biochemically related vibrational modes that serve to detect bacterial presence and allow for the successful screening for these microbial organisms. These are outlined and summarized in Table 1. Two-dimensional co-distribution and two-dimensional correlation spectroscopies have been employed to establish the relationship between the key signature peaks, and provide evidence of bacterial presence in a sample without a priori knowledge as to the source or conditions of the sample (FIGs.7A-B, 8A-D, 9A-B, 10A-D) for Gram negative bacteria Escherichia coli and Pseudomonas aeruginosa. The system and method take advantage of the different growth rates observed for different microbial organisms when compared to other hosts cells that have been genetically engineered for the production of therapeutic proteins. The 2D-CDS asynchronous analysis provides the overall distribution of key signature peaks for populations of cells in the samples examined, allowing for the discrimination of samples comprised of bacterial presence (FIGs.7A-B and 9A-B) for E. coli and P. aeruginosa, respectively. Moreover, the method does not require an incubation period to examine a larger population of follow-on generations of the microbial organism. Similarly, the 2D-COS synchronous and asynchronous plots are shown in FIGs.8A-D and 10A-10, for E. coli and P. aeruginosa, respectively. The 2D-COS cross peak analysis provides the relationship between the key signature peaks and allows for the determination of the multiple components within the sample and can be used to ascertain the biochemical state of the host and bacterial contaminant. Further analysis of the key signature peak ratios provides information regarding the extent of bacterial and endotoxin contamination within each sample of the slide cell array. Before looking for key signature peaks, the system may perform a baseline correction to spectral data for one or more regions of interest. In particular, there is typically scatter, which causes a deviation of the base of the spectrum, which may appear as an arc. The baseline correction removes this scatter. The baseline correction may be a spline correction that removes the arc, and can do so by using spline points at, for example, 1770, 1482, 1350, 1188, and 1006 cm ^-1 . Key signature peak ratios provide the advantage of not having the concentration dependence resulting in accurate determination regarding both bacterial and endotoxin presence in the sample at any given time. A summary of the key signature peaks and their assignments are shown in Table 1:

Referring to Table 1 above, for bacterial presence the signature peaks are: methylene deformation modes ( ^(CH2)) at 1452 cm ^-1 mainly from the lipid acyl chains in membranes and non-polar side chains in proteins; the stretching carbonyl mode in carboxylates ( ^(C=O), COO-) at 1404 cm ^-1 from carbohydrates; phosphodiester bond is associated with two stretching modes ( ^a(PO2)) at 1236 cm ^-1 from phospholipid membrane, of which only one is observed in bacteria; and a carbohydrate ring stretching vibration ( ^a(C-O-C)) at 1084 cm ^-1 . For the endotoxin all of the peaks listed for bacteria are also considered and observed at higher levels as well as the second phosphodiester stretching mode ( ^s(PO2)) at 1156 cm ^-1 and carbohydrates at 1115 cm ^-1 and 1038 cm ^-1 observed due to the shedding of the lipolysaccharide (LPS) upon death of the Gram-negative bacteria. The peak intensity ratios are used to for quantitative analysis. The system and method provide several bacterial and endotoxin screening advantages: (1) no incubation period is required allowing for a real-time analysis; (2) the evaluation of liquid cultures/samples by using amide II/amide I peak intensity ratio to ensure reproducibility of the analysis; (3) the use of a fixed path-length slide array allows for comparability assessment among different samples; (4) the evaluation of liquid cultures at a fixed path-length also allows for quantitative analysis and orthogonality between optical techniques; (5) analysis of multiple peaks to assess bacterial presence allows for high degree of confidence in the determinations; (6) complex samples comprised of multiple organisms may also be analyzed. The above-mentioned advantages are also valid for the direct endotoxin detection. FIG.1 is an illustration of a summary or overview of Quantum Cascade Laser (QCL) transmission microscope hyperspectral image analysis for a slide cell array. FIG.1 shows a   hyperspectral image (1) of a sample well, containing a sample, of a slide cell array at 4.3 ^m spatial resolution. FIG.1 further shows an enhanced S/N QCL IRM overlay (2) and orthogonality (3) and 2D correlation analyses (4,5,6). In FIG.1, (4) shows 2D-CDS and (5, 6) show 2D-COS of the hyperspectral image of the sample. The analysis ensures the real-time and direct monitoring of bacterial and endotoxin presence. For direct monitoring of bacterial presence, with orthogonality to other assessments, such as OD 600nm, pellet-cell volume (PCV), and turbidity. The identification of signature peaks and their correlations allow for the determination of the presence of the bacterial/endotoxin contaminant. Furthermore, the slide cell array allows for comparability assessment and quantitative analysis between samples. Finally, positive and negative controls are also included in the slide cell array. For example, a positive control sample having the bacterial/endotoxin control standard may be included to provide a sample well that should indicate a positive result. Similarly, a negative control sample may be included in a sample well that is known to not include any bacteria/endotoxin, such as water, that should provide a negative result (that is, a result with no key signature peaks observed). In the system and method for detecting endotoxin, one or more of the samples or sample wells may be spiked with Control Standard Endotoxin (CSE). For example, one or more samples or sample wells may be spiked with 15 EU/ML of CSE. FIG.2 is an illustration of a real-time bacterial and endotoxin detection flow chart – acceptance criteria to ensure reproducibility and Go No-Go decision making. The amide II band associated with protein side chain modes and the amide I band comprised of overlapping conformationally sensitive peptide bond modes and the H2O bending vibration intensity ratios serve to access relative aqueous content in the sample to ensure reproducibility of sample preparation. A threshold value has been defined for acceptance of the QCL infrared microscopy (IRM) data prior to commitment of complete spectral data analysis. As shown in FIG.2, with respect to bacteria detection, the process may include acquiring a plurality of hyperspectral images of the samples in the slide cell array having a plurality of sample wells over time. The slide cell array may include liquid cultures and samples, and hyperspectral images of these liquid cultures and samples are obtained. The hyperspectral images may be acquired using a quantum cascade laser microscope, or any other suitable method to obtain spectral data such as a Raman amplifier or LIDAR. For example, the spectral images of the sample may be acquired using a quantum cascade laser microscope, wherein the slide provides a fixed path length of between 3-5µm. The   hyperspectral images are then analyzed, and regions of interest (ROIs) within the field of view are identified and exported as spectral data for at least a first spectral image. For example, spectral data for the regions of interest may be exported as QCL IRM data within the region of 1780 – 1000 cm ^-1 . A baseline correction may also be applied to the spectral data for the ROIs. This baseline-corrected spectral data for the regions of interest may then the analyzed to determine peak intensities of the data at wavenumbers corresponding to the amide I and amide II bands. Peak intensities for other vibrational modes, as shown in Table 1 herein, are also determined. A ratio of the peak intensity of the amide II band to the peak intensity of the amide I band is then determined, and compared to a threshold limit. The predetermined threshold for the ratio of amide II to amide I peak intensity may be, for example, 0.4 to 0.6. If the amide II to amide I band ratio do not meet the threshold limit, then the process is ended as the data is considered insufficient for further spectral analysis. Similar, if there is not a linear dependence between related samples, then the acquired data is also considered insufficient for further spectral analysis. With respect to endotoxin detection, the system and method can also be used to detect the presence of endotoxins within a sample, such as samples in the wells of a slide. A first amount of a sample may be provided in at least one sample well. A second amount of the sample may be treated to kill any microbes that may be present, and this second amount of sample may be provided in at least one second sample well. The second amount of sample may be treated to kill any microbes by, for example, heat treating, ultraviolet light, or chemical treating. Spectral images of the first sample amount and second sample amount are then acquired. In particular, a first spectra image of the first sample amount is acquire and a second sample image of the second sample amount is also acquired. These spectral images may be acquired using a quantum cascade laser microscope, or any other suitable method to obtain spectral data such as a Raman amplifier or LIDAR. For example, the spectral images of the sample may be acquired using a quantum cascade laser microscope, wherein the slide provides a fixed path length of between 3-5µm. Regions of interest are then identified and selected in each of the first and second spectral images. These regions of interest may be the entire field of view, or may be select portions of the field of view of the images. First spectral data for the region of interest in the first spectral image is obtained, and second spectral data for the region of interest in the second spectral image is also obtained. The first and second spectral data may then be baseline corrected. This baseline-corrected spectral data for the regions of interest may then the analyzed to determine peak intensities of the data at wavenumbers corresponding to the amide I and amide II bands. Peak intensities for other   vibrational modes, as shown in Table 1 herein, are also determined. A ratio of the peak intensity of the amide II band to the peak intensity of the amide I band is then determined, and compared to a threshold limit. The predetermined threshold for the ratio of amide II to amide I peak intensity may be, for example, greater than about 0.35. If the amide II to amide I band ratio do not meet the threshold limit, then the process is ended as the data is considered insufficient for further spectral analysis. Similar, if there is not a linear dependence between related samples, then the acquired data is also considered insufficient for further spectral analysis. If the spectral data from the acquired spectral images in is considered sufficient for further spectral analysis, the process can continue to perform detection of bacteria and endotoxins within the liquid cultures and samples. FIG.3 is an illustration of a real-time bacterial and direct endotoxin detection flow chart – application of multiple algorithms. The actual quantitative analysis is performed after the 2D-CDS and 2D-COS analyses of the samples in the array, resulting in an accurate, reproducible analysis. If the presence of Gram negative bacteria is confirmed, then evaluation of endotoxin presence follows. As shown in FIG.3, the process may include generating two-dimensional 2D-CDS asynchronous plots from the spectral data. Cross peaks in the 2D-CDS asynchronous plots are analyzed, and a determination is made as to whether there are correlations involving key signature cross-peaks. If there are no observed correlations between the key signature cross- peaks, the process may be terminated. If correlations between cross-peaks are observed, 2D- COS synchronous and asynchronous plots may be generated for the spectral data. In particular, peak intensities of the baseline corrected data at wavenumbers corresponding to the vibrational modes identified in Table 1 may be determined. For example, the peak intensities for aet least two of at least two of carbohydrates (C-O and COO-), and membrane associated vibrational modes (PO ₂ and CH ₂ ) may be determined. One or more peak intensity ratios for peak intensities of the one or more vibrational modes compared to one or more other vibrational modes are determined. The peak intensity ratios are then analyzed with respect to a control, such as water. The peak intensities are analyzed to determine whether they increase relative to the control. If one or more of the peak intensity ratios increases relative to the control, the process indicates the presence of a microorganism, such as bacteria. The process may further include acquiring at least one second spectral image of one or more samples using a quantum cascade laser microscope, wherein the at least one second spectral image is acquired for the sample at a different concentration or after the application   of heat to the sample. Spectral data for the at least one second spectral image, which may be a single spectral image or a plurality of spectral images, are then obtained. Two-dimensional co-distribution analysis to the obtained spectral data for the at least one second spectral image to generate a co-distribution plot for the sample. The co-distribution plot is analyzed, and cross-peaks are identified within the plot. The cross-peaks are analyzed to determine if the co-distribution plot includes at least one cross-peak associated with the presence of a microorganism in the sample. If a cross-peak associated with the presence of a microorganism is detected in the plot, then the presence of a microorganism in the sample is confirmed. Moreover, if the cross-peaks in the co-distribution plot are at expected positions, this indicates that the sample is homogeneous. The process may further include applying two-dimensional correlation (2D-COS) analysis to the spectral data from the regions of interest for the acquired hyperspectral images. The 2D-COS analysis may be applied to generate a synchronous correlation plot and an asynchronous correlation plot. Once the plots are generated, cross peaks that correlate with the presence of the microorganism can be identified in the synchronous correlation plot. With respect to endotoxin detection, the first and second baseline-corrected spectral data corresponding to the images for the first and second sample amounts is further analyzed. Peak intensities of the baseline corrected data at wavenumbers corresponding to the vibrational modes identified in Table 1 may be determined for the first spectral data and the second spectral data. For example, the peak intensities for at least two of at least two of carbohydrates (C-O and COO), and membrane associated vibrational modes (PO ₂ and CH ₂ ) may be determined for the first spectral data and second spectral data. One or more peak intensity ratios for peak intensities of the one or more vibrational modes compared to one or more other vibrational modes are determined for the first spectral data. And, one or more peak intensity ratios for peak intensities of the one or more vibrational modes compared to one or more other vibrational modes are determined for the second spectral data. The peak intensity ratios for the first spectral data are then compared to the peak intensity ratios for the second spectral data. Test ratios are generated from these comparisons of the peak intensity ratios of the first spectral data to the peak intensity ratios for the second spectral data. When one or more of the test ratios is within a predetermined threshold, this indicates the presence of an endotoxin. The predetermined test ratio threshold may be less than or equal to about 0.6 or greater than or equal to about 1.4. The method of endotoxin detection may further include acquiring at least one third spectral image, which may be at least one hyperspectral image. This at least one third spectral   image may be acquired for the sample at a different concentration or after the application of heat to the sample. The method may then include obtaining spectral data for the at least one third spectral image, and applying two-dimensional co-distribution analysis to the obtained spectral data for the at least one third spectral image to generate a co-distribution plot for the sample. The co-distribution plot is then analyzed and a determination made as to whether it includes at least one cross peak associated with the presence of a microorganism or endotoxin in the sample. When a cross peak associated with the presence of a microorganism or endotoxin in included in the plot, the presence of the endotoxin in the sample is confirmed. The method may further comprise analyzing the co-distribution plot, and determining the sample is homogeneous when peaks in the co-distribution plot are at expected positions. In addition, two- dimensional correlation analysis may be applied to the obtained spectral data for the at least one third spectral image to generate synchronous and asynchronous correlation plots for the sample. These plots can be analyzed, and a determination made as to whether the synchronous and asynchronous correlation plots include at least two cross peaks associated with the presence of the microorganism or endotoxin in the sample. When the at least two cross peaks are included in the synchronous and asynchronous correlation plots, this confirms the indication of the presence of the microorganism or endotoxin in the sample. FIGs.4A and 4B are an example of a typical hyperspectral image and associated QCLM spectra for a bacterial sample acquired using a QCLM, respectively. The spatial resolution was 4.3 ^m. The QCL IR spectral data shown within the spectral region of 1780 – 1000 cm ^-1 demonstrates the homogeneous distribution of the bacterial sample. FIG.4A shows typical hyperspectral image. Also highlighted on the hyperspectral image in FIG.4A are six regions of interest (ROIs) as squares. FIG.4B shows the corresponding QCLM spectra for the field of view and six ROIs of FIG.4A, demonstrating homogeneity of the bacterial sample distribution. The biochemical composition of bacteria shown through key signature peaks representing different vibrational modes characteristic of these components. The vibrational mode characteristics of these components are identified above in Table 1. FIG.5 shows linearity and orthogonality for optical density (OD) 600 nm and QCLM for three Gram negative bacterial species obtained from the American Type Culture Collection (ATCC): (closed square) Escherichia coli (ATCC CRM-8739), (open circle) Pseudomonas aeruginosa (ATCC CRM-9027) and (open up triangle) Brevundimona diminuta (ATCC 19146) determinations for all of the signature peaks proposed are shown. This approach allowed for method qualification and validation. Each plot is for a different   key signature peak (wavenumber). Each key signature peak is evaluated on its own for linearity and orthogonality with the optical density (OD) 600nm. Each symbol in the plot is for a different bacterial strain. FIG.6 shows linearity and orthogonality for optical density (OD) 600 nm and QCLM for three Gram positive bacterial species for three bacterial species obtained from the American Type Culture Collection (ATCC): (open diamond) Staphylococcus aereus (ATCC CRM-6538), (plus sign) Clostridium Sporogenes (ATCC CRM-11437), and (closed up triangle) Cutibacterium acnes (ATCC 11827 determinations for all of the signature peaks proposed are shown. This approach allowed for method qualification and validation. Each plot is for a different key signature peak (wavenumber). Each key signature peak is evaluated on its own for linearity and orthogonality with the optical density (OD) 600nm. Each symbol in the plot is for a different bacterial strain. FIG.7A shows a 2D co-distribution asynchronous plot for E. coli within the spectral region of 1780 – 1000 cm ^-1 , and FIG.7B shows a 2D co-distribution asynchronous plot for E. coli within the spectral region of 1482 -– 1000 cm ^-1 . Also, the intensity changes can be defined by the grey scale bar shown at the right side of FIGs.7A and 7B. This type of analysis is ideal for the detection of bacterial presence in a sample. Specific signature cross peaks and their correlations provide evidence of the bacterial presence. FIGs.8A-8D show 2D-COS plots for E. coli within the spectral region of (top row – FIGs.8A and 8B) 1780 – 1000 cm ^-1 and (bottom row – FIGs.8C and 8D) 1482 - 1000 cm ^-1 , which are used to investigate further the bacterial contamination within the sample. FIGs.8A and 8C are the synchronous plots and FIGs.8B and 8D are the asynchronous plots. Also, the intensity changes can be defined by the grey scale bar shown at the right-hand side of FIGs. 8A-8D. The investigation may include, but is not limited to, the classification of the bacterial contamination, the additional components within the sample and the biochemical state of the sample. FIG.9A shows a 2D co-distribution asynchronous plot for Pseudomonas aeruginosa within the spectral region of 1780 – 1000 cm ^-1 and FIG.9B shows a 2D co-distribution asynchronous plot for Pseudomonas aeruginosa within the spectral region of 1482 -– 1000 cm ^-1 . Also, the intensitsy changes can be defined by the grey scale bar at the right side of each of FIG.9A and 9B. This type of analysis is ideal for the detection of bacterial presence in a sample. Specific signature cross peaks and their correlations provide evidence of the bacterial presence.   FIGs.10A-10D show 2D-COS plots for Pseudomonas aeruginosa within the spectral region of (top row – FIGs.10A and 10B) 1780 – 1000 cm ^-1 and (bottom row – FIGs.10C and 10D) 1482 - 1000 cm ^-1 , used to investigate further the bacterial contamination within the sample. FIGs.10A and 10C are the synchronous plots and FIGs.10B and 10D are the asynchronous plots. Also, the intensity changes can be defined by the grey scale bar on the right side of each of FIGs.10A-10D. The investigation may include, but is not limited, to the classification of the bacterial contamination, the additional components within the sample and the biochemical state of the sample. Monitoring Bacterial Cell Growth The system and method may also be used to evaluated and monitor bacteria cell growth. By evaluating the absorptivity at key signature peaks over time, different phases of bacteria cell growth and death can be monitored and evaluated, including a lag phase where there is no growth, a growth phase, a stationary phase where the bacteria are no longer growing at a significant rate, and a death phase where the bacteria cells die. To conduct this monitoring, a slide comprising a plurality of sample wells may be provided, and samples included in one or more of the wells. Spectral images of the samples in the wells are acquired as a function of time. The spectral images may be hyperspectral images. The spectral images may be acquired by any suitable method such as, for example, a quantum cascade laser microscope, a Raman amplifier, LIDAR, infrared spectroscopy, or combinations thereof. After the spectral images are obtained, spectral data for one or more regions of interest in the spectral images are acquired. The regions of interest may be the entire field of view, or portions thereof. Key signature peaks for the spectral data are determined. For example, with respect to FIG.11, a Gram negative bacteria, Pseudomonas aeruginosa, was used to evaluate the sensitivity of four key signature peaks (1085, 1237, 1404, and 1454 cm ^-1 ) to monitor the lag, growth, stationary and death phases of a growth curve. The growth curve may illustrate the absorbance over time. As seen in FIG.11, the lag phase would have little to no absorbance, while the growth phase exhibits increasing absorbance with time. This increasing absorbance may be at a rate comparable to the doubling time known for the bacterial strain. A selection of the time points within the growth phase may also evaluated via turbidity assessment using the McFarland standards (1-4) to establish equivalency. The stationary phase is characteristic of high absorbance with little to no variability, followed by the death phase where a decrease in absorbance is observed. All key signature peaks proved to be valid for the monitoring of bacterial growth, as demonstrated by FIGs.   12A-B and 13A-B. Time point spectral image samples were collected for analysis initially every two hours (during lag phase) followed by every hour during all subsequent phases of the cell culture. FIG.12A illustrates the growth curve for the signature peak at 1085cm ^-1 over a time from 12-25 hours. A lag period is seen between 12-16 hours, followed by a growth period from 16-22 hours, a stationary period from 22-23 hours, and then a death period from 24-25 hours. FIGs.12B, 13A, and 13B illustrate the same growth curve with observable lag, growth, stationary, and death phases for signature peaks at 1237 cm ^-1 , 1404 cm ^-1 , and 1454 cm ^-1 , respectively. Differences in sensitivity are based on the absorptivity for each of the functional groups that represent a different biochemical component within the bacteria. These components may correspond to the different key signature peaks assigned to different components as shown in Table 1. For example, higher absorptivity was observed for the 1085 cm ^-1 signature peak assigned to the carbohydrates. As illustrated in FIG.14, a slide cell array having a plurality of sample wells may be provided such that the key peak absorptivity can be determined for a sample, as described above. The slide array may include three or more rows of sample wells. The slide cell array shown in FIG.14 allows for triplicate measurements of each time point sampled, as the sample can be provided into three wells at each time point, t1 to t7, as shown. The time t1-t7 may correspond to, for example, 12-21 hours. The analysis can be performed using a small amount of sample in each well, such as 2 ^L of sample per well. Moreover, turbidity of the sample can be observed as a function of time (t1-t7 for 12 – 21 hours, respectively). A summary of the key signature peak absorbances is shown in Table 2 below:

  While the above example from FIGs.11, 12A-B, 13A-B and14 was described with respect to Pseudomonas aeruginosa, the system sand method can be implemented to evaluate multiple Gram negative and positive bacteria for all bacterial strains. The bacterial strains may be evaluated using the same key signature peaks for the determination of the bacterial presence in the cell culture, or particular key signature peaks can be identified for different bacterial strains. The system and method allow for monitoring QCL IRM absorbance of the key signature peaks, which is then used to determine bacterial cell growth over time. Factors that are known to influence bacterial growth are bacterial strain classification, the media, and the conditions to which the bacteria are subject to during incubation. Monitoring the absorbance of the key peaks allows monitoring of impact of different conditions on bacteria growth during incubation. The equivalency to CFU/mL and lead time of detection can be determined. Endotoxin Detection Using 2T2D Correlation Analysis The system and method may further include using two-trace two-dimensional (2T2D) correlation analysis for the determination of endotoxin presence. The use of 2T2D analysis to detect endotoxin requires only the spectral data for a sample and spectral data for a reference to conduct the evaluation. Spectral images, which may be hyperspectral images, are acquired as described above. These spectral images may be acquired at a particular temperature, such as 37 ^o C, and no tags or labels need to be used to identify the endotoxin,   which could be for example lipopolysaccharide (LPS). Rather, 2T2D correlation analysis is applied to the spectral data. From a pair of spectra a two trace two-dimensional correlation can be applied to generate a synchronous spectrum Φ _ଶ and asynchronous spectrum Ψ , _ଶ The synchronous spectrum ^ and asynchronous spectrum Ψ , _ଶ are given by: The first, original spectra, corresponds to the sample and the second spectra, , corresponds to the reference, respectively. A 2T2D correlation coefficient is then _{applied to the synchronous evaluation, and a disrelation coefficient o} the asynchronous evaluation resulting in the scaled version of the 2T2D correlation spectra, as given by: This indicates the complementarity nature of the quantities. Two contour plots are generated. These are the synchronous ^ plot where dominant spectral components of the mixture: ^ d are observed. The diagonal is comprised of auto peaks where ൌ The cross peaks are always positive. The second plot is the asynchronous plot, which is more informative. In the case then the intensity contribution of the functional group is from ^ corresponding to the first component being more abundant. In the case ^ ^ then the intensity contribution of the functional group is from ^, therefore, the second component of the mixture is more abundant.   Also, this indicates that peaks of the same intensity and sign correspond to the same component within the mixture. FIGs.15A and 15B show an example of 2T2D analysis within the spectral region of 1485 – 1000 cm ^-1 analysis for Pseudomonas aeruginosa a Gram negative bacteria. FIG.15A illustrates a synchronous plot and 15B illustrates an asynchronous plot. In FIG.15B, the positive and negative signs identify the locations of the cross peaks within the plot, with a positive sign indicating that the peak is positive and a negative sign indicating the cross peak is negative. For the synchronous plot all of the peaks are positive and the peaks on the diagonal are known as the auto peaks and represent the overall magnitude of the spectral components in the spectra analyzed. The asynchronous plot is highly informative in that it allows for the identification of the spectral components of the mixture present in the sample. As can be seen, the peaks inverted from positive to negative as the cell changed with time. The phase differentiation is attributable to each component of the mixture. Two different sets of cross peaks can be observed suggesting a two-component mixture. The components comprised of positive sign cross peaks (1404,1084), (1404,1115), (1454, 1115), (1084, 1060), (1156, 1084), (1235,1115), (1454, 1235), (1404,1235) and (1454, 1404) that contained a significant contribution of ^(CH ₂ ) at 1454 cm ^-1 , v(C=O) of COO- at 1404 cm ^-1 , phosphodiester ^s(PO2) and ^a(PO2) at 1156 and 1235 cm ^-1 , respectively; and carbohydrate ring vibrational modes at 1115, 1084 and 1060 cm ^-1 suggesting the endotoxin component due to the well-known presence of these chemical groups in lipopolysaccharides. The second component comprised of negative cross peaks (1365, 1084), (1235,1084), (1156, 1084) containing amide III band at 1365 cm ^-1 , phosphodiester v _s (PO ₂ ) and v _a (PO ₂ ) at 1156 and 1235 cm ^-1 and carbohydrate associated vibrational mode at 1084 cm ^-1 suggesting they are associated with the bacteria. Also, the observed correlations between the positive and negative cross peaks supports the Gram negative classification of the bacteria due to the correlations with the endotoxin. The weighted difference spectra for the spectral data reflected in FIGs.15A and 15B are shown in FIGs.16A-B, providing the spectral evidence of the components of the mixture within the P. aeruginosa cell culture sample. FIG.16A shows a weighted difference spectra for the endotoxin component, and FIG.16B shows a weighted difference spectra for P. aeruginosa cells. The weighted spectral difference for endotoxin is very similar to that of control standard endotoxin (CSE). And, the resulting weighted difference spectrum for the   bacterial component has the minimal components necessary to identify it as a separate component within the mixture. Bacterial and Endotoxin Detection During Host Cell Bioprocessing The system and method described herein may also be used to detect bacteria and endotoxin during host cell bioprocessing. The host cell may be, for example, a mammalian cell, insect cell, fungi cell, or plant cell. This may be performed, for example, during upstream bioprocessing at each point prior to every cell culture expansion limiting tandem contamination within upstream bioprocessing. Furthermore, the potential risk for endotoxin contamination during downstream processing may also be averted. These measures lead to ensuring speed to market and become effective cost-efficient measures that ensure regulatory compliance in manufacturing. The assessment of contamination is verified by the application of multiple correlation algorithms including, but not limited to: 2D-CDS, 2D-COS and 2T2D spectroscopies for baseline corrected spectral data. The asynchronous two-dimensional co- distribution spectroscopy provides the changes in distribution of cellular population in the sample being evaluated as a function of time. For example, FIGs.17A-17B illustrate co- distribution plots of mammalian cell culture samples harvested as a function of time during experimental growth (FIG.17A) and experimental simulated dying conditions (FIG.17B). In the co-distribution plots, the arrows point to the location of key cross peaks, and the positive and negative signs next to the arrows indicate whether the peak is positive or negative. Positive peaks indicate that the host cells are growing, while negative peaks indicate that the host culture cell is dying, such as if they are being challenged by a microbial contaminant (i.e., bioburden). In general, positive peaks are indicative of growing organisms, and here would reflect the host cells growing. Where these peaks are negative instead of positive, it can be determined that the host cells are dying rather than growing. In the case where the sample includes both a desired host cell and a microbial contaminant, the growth state of both the host cell and microbial contaminant can be determined by looking at the key cross peaks based on their different growth rates. FIG.18 is schematic diagram of the workflow for the evaluation of a host cell, such as mammalian cell, culture sample being spiked with E. coli. As shown in FIG.18, a host cell culture, such as a mammalian cell culture, can be provided in a bioreactor. Sample amounts of the cell culture can then be extracted, and controlled amounts of E. coli culture, based on McFarland Standards, added to the sample. Spectral images of the sample spiked with E. coli, as well as samples of the host cell culture without any E. coli, can then be obtained for analysis. In particular, 2D-CDS, 2D-COS and 2T2D analysis can be performed as described   herein. As part of the analysis, the absorbance of key signature peaks may be evaluated. The absorbance of the key signature peaks as a function of E. coli being spiked unto a fixed time point host cell culture sample resulting in multiple samples under increasing presence of spiked E. coli is illustrated in FIGs.19A-D. The data in FIGs.19A-D may be obtained from a quantum cascade laser microscope (QCLM). In particular, FIG.19A illustrates the absorbance of the key signature peak of 1094 cm ^-1 while FIG.19B illustrates the absorbance of the key signature peak of 1235 cm ^-1 , FIG.19C illustrates the absorbance of the key signature peak of 1454 cm ^-1 and FIG.19D illustrates the absorbance of the key signature peak of 1404 cm ^-1 . As can be seen, as more E. coli is spiked into the sample, the absorbance increases. FIG.19E illustrates the amount of spiked E. coli included in the sample. A summary of the QCLM average key signature peak absorbances for the mammalian cell culture spike with E. coli is shown in Table 3 below: FIGs.20A and 20B show the co-distribution analysis for an experimental simulation designed to include the scenario where the host cell culture is dying, while increasing E. coli cells/mL are being spiked. FIG.20A illustrates a 2D-CDS asynchronous correlation plot for a mammalian cell culture under dying conditions in the spectral region of 1485 – 1000 cm ^-1 , and FIG.20B illustrates a 2D-CDS asynchronous plot for a mammalian cell culture in the presence of E. coli (spiked) in the spectral region of 1485 – 1000 cm ^-1 . The mammalian cells may be spiked with E. coli cells at 25 ^o C. Cross peaks in the plots are identified with plus and minus signs, indicating whether the peaks are positive or negative. As can be seen, the sign of the cross peak changes in the presence of the spiked E. coli when compared to the mammalian cell culture under dying conditions. Cross peaks attributable to each component of the sample mixture may therefore be identified. Details of the cellular dynamics of the cell culture sample can be examined using the two-dimensional correlation asynchronous plot.   In addition, 2D-COS plots can be used to identify the key signature peaks and can also be used to determine the state of the host cell in the presence of the growing bacterial contaminant. For example, FIGs.21A-D show synchronous and asynchronous 2D-COS plots for host control cells and host cells spiked with E. coli. In particular, FIG.21A shows a 2D- COS synchronous plot for experimentally simulated dying mammalian cells for experimental data within the spectral region 1485 – 1000 cm ^-1 . FIG.21B shows a 2D-COS synchronous plot for experimentally simulated dying mammalian cells in the presence of increasing E. coli cells for experimental data within the spectral region 1485 – 1000 cm ^-1 . FIG.21C shows a 2D-COS asynchronous plot for experimentally simulated dying mammalian cells for experimental data within the spectral region 1485 – 1000 cm ^-1 . FIG.21D shows a 2D-COS asynchronous plot for experimentally simulated dying mammalian cells in the presence of increasing E. coli cells for experimental data within the spectral region 1485 – 1000 cm ^-1 . In the synchronous plots, the correlations between the key signature peaks are evident. The identification of the key cross peaks in the plots allows for the identification of the different components in the mixture solution. That is, when certain cross peaks are identified in the plots, this is an indication of the presence of components such as bacteria in the solution. For example, different components, such as bacteria, in the mixture will cause different cross peaks with different signs as compared to control host cells under dying conditions. When these particular peaks are observed, the sample can be considered contaminated. This observation is not limited to bacterial strains, but can also be performed for viral and Fungi contaminant. Similarly for the host cell, the approach is not limited by host cell type, such as insect or Pichia pastoris. This is due to the advantage of these correlation algorithms that when applied the differentiation is based on the cells different growth rates between the host and the contaminating organism. Direct Determination of Endotoxin Contamination in Host Cell Cultures Acquiring and analyzing spectral images, such as hyperspectral images, can allow direct detection of an endotoxin contamination in a host cell culture. When a host cell culture is contaminated, 2T2D analysis of the spectral images can indicate the presence of key cross peaks associated with an endotoxin. When these cross peaks are present, the host cell culture can be determined to be contaminated. As discussed above, the spectral images may be hyperspectral images obtained using a quantum cascade laser microscope. The images may also be obtained using any vibrational spectroscopy technique, such as a Raman amplifier or LIDAR. A typical hyperspectral image of mammalian cells spiked with control standard endotoxin is shown in FIG.22A, and a typical QCLM spectrum for the same mammalian   cells spiked with control standard endotoxin is shown in FIG.22B.2T2D analysis was performed on cells, and demonstrates that a determination of endotoxin presence in the mammalian cell culture which was spiked with control standard endotoxin (CSE) can be made. Thus, endotoxin in cell cultures can be detected using the system and method described herein. FIGs.23A and 2B represent the synchronous and asynchronous plots, respectively, for 2T2D analysis of spectral data for a mammalian cell culture that was spiked with CSE. As described above, the asynchronous plot represents the major contributions from the components of the mixture present in the cell culture sample. In the asynchronous plot shown in FIG.23B, cross peaks are identified with plus and minus signals, indicating the sign of each peak. The positive cross peaks are attributable to the endotoxin, while the negative peaks are attributable to the proteinaceous nature of the mammalian cells. Specifically, the main positive cross peaks: (1630, 1535), (1630, 1237) and (1630, 1175) all attributable to the endotoxin component of the mixture such as the glucosamine amide at 1630 cm ^-1 , phosphodiester stretching at 1237 cm ^-1 and carbohydrate ring mode at 1175 cm ^-1 . While the main negative cross peaks: (1660, 1636), (1660, 1076), (1660, 1043), (1545, 1076) are associated with the proteinaceous and cellular components of the mammalian cell such as amide I band at 1660 and 1636 cm ^-1 , carbohydrate moiety 1545, 1076, and 1043 cm ^-1 . The asynchronous plot also provides enhanced spectral resolution and correlations that allow for distinguishing between spectral components within a mixture. Thus, by analyzing the cross peaks in the asynchronous plot, the presence of endotoxin in a sample can be detected. The weighted spectral difference for each component of the mixture is shown in FIGs. 24A and 24B, with FIG.24A being for the endotoxin and FIG.24B being for the mammalian cells. For the endotoxin component comprised of the lipid carbonyl at 1726 cm ^-1 , the glucosamine amide component at 1630 cm ^-1 , the methylene in-plane bending 1454 cm ^-1 , carbohydrate stretching (C=O), COO-, the phosphodiester asymmetric and symmetric stretches at 1237 and ~1156 cm ^-1 , respectively; and carbohydrates vibrational modes at 1115, ~1080 and ~1060 cm ^-1 of the endotoxin. The mammalian cells are illustrated as being the amide I, II and III bands at 1645, 1550 and ~1365 cm ^-1 , respectively. These are the main spectral features that are distinguishable from both spectral components. As such, the spectral features can be used to detect the presence of endotoxin in samples. In the system and method described herein, spectral images serve as a historical image of the sampling event. The system and method allow for real-time detection capabilities after a short sample preparation step. The determination can inform the state of the host cell   culture in relation to its bacterial contaminant. The results are accurate and reproducible within the limits of detection and quantitation. Multiple points of sampling may also be implemented to ensure extent of bacterial contamination for forensic investigation in the manufacturing setting. In addition, direct endotoxin detection is based on the comparative analysis of a reference standard against a sample, and evidence of the mixture components within the sample can be ascertained. When the endotoxin presence is confirmed then the classification of the bacterial contamination can be defined as Gram Negative. The system and method may be implemented during the upstream processing or downstream processing of a manufacturing or other process. There may be multiple points at which to take samples for bacterial and endotoxin screening. For example, as shown in FIG. 25, an overall biopharma process may include steps or line-sampling points of filtered air inoculum and media feed, perfusion bioreactor, filtration, affinity column, low pH viral inactivation, and multiple chromatographic filtration steps. The potential for improved bacterial and endotoxin screening in biopharma manufacturing operations both upstream and downstream processes as an at-line platform implementation allowing for multiple sampling points for evaluation of both bacterial and endotoxin screening.