Hyperspectral imaging for red blood cell membrane oxidative status

Abstract

Various examples are provided related to hyperspectral imaging (HSI) of red blood cell (RBC) membranes. In one example, a method for determining whether a subject has a condition or is predisposed for developing the condition includes using a visible near infrared (VNIR) hyperspectral imaging system to obtain a hyperspectral image (HSI) of a RBC membrane of the subject over a range of wavelengths to obtain HSI spectra and determining whether the HSI spectra is indicative of oxidative stress in the RBC membrane, where the presence of the oxidative stress in the RBC membrane can indicate the subject has the condition or is predisposed for developing the condition. The wavelengths used to obtain the HSI spectra are in the VNIR range. The method is non-invasive and does not comprise administration of dyes for detection of the oxidative stress in the RBC membrane.

Description

Docket: 220312-2030 HYPERSPECTRAL IMAGING FOR RED BLOOD CELL MEMBRANE OXIDATIVE STATUS CROSS REFERENCE TO RELATED APPLICATIONS

[0001] This application claims priority to, and the benefit of, U.S. provisional application entitled “Hyperspectral Imaging for Red Blood Cell Membrane Oxidative Status” having serial no.63 / 612,633, filed December 20, 2023, which is hereby incorporated by reference in its entirety. BACKGROUND

[0002] The principal function of red blood cells -or erythrocytes- is to carry oxygen from the lungs and nutrients in order to deliver them to the tissues of our entire body. Recent advances have evidenced that red blood cells (RBCs) act also as interorgan communication systems with additional functions, such as the participation in control of systemic nitric oxide metabolism, redox regulation, blood rheology, and viscosity. In RBC the molecular analysis focused on hemoglobin, which has been studied for its reactivity and participation to multiple chemical and enzymatic redox processes (nitrite reductase, NO dioxygenase, monooxygenase, alkylhydroperoxidase, esterase, lipoxygenase); nitric oxide metabolism; metabolic reprogramming; pH regulation and maintaining redox balance. Work on the visible and near infrared region of hemoglobin absorption began in 1942. Methodologies that can be used for RBC detection without the use of any biomarker or the need for membrane destruction processes include infrared spectroscopy (IR), atomic force microscopy (AFM), Fourier-transform spectroscopy (FT-spectroscopy). SUMMARY

[0001] Aspects of the present disclosure are related to hyperspectral imaging (HSI) of red blood cell (RBC) membranes. In one aspect, among others, a method for determining whether a subject has a condition or is predisposed for developing the condition comprisesDocket: 220312-2030 using a visible near infrared (VNIR) hyperspectral imaging system to obtain a hyperspectral image (HSI) of a red blood cell (RBC) membrane of the subject over a range of wavelengths to obtain HSI spectra and determining whether the HSI spectra is indicative of oxidative stress in the RBC membrane, wherein the presence of the oxidative stress in the RBC membrane indicates the subject has the condition or is predisposed for developing the condition, wherein the wavelengths used to obtain the HSI spectra are in the VNIR range, wherein the method is non-invasive and does not comprise administration of dyes for detection of the oxidative stress in the RBC membrane. In one or more aspects, the condition can be autism.

[0002] In various aspects, the HSI spectra of the RBC membrane can be compared to control reference HSI spectra to determine delta variations characterizing a status of the RBC membrane, the status associated with the oxidative stress of the RBC membrane. The HSI spectra can comprise 8 endmember spectra of the RBC membrane. The delta variations can comprise increases or decreases in defined wavelength ranges in the endmember spectra. The status of the RBC membrane associated with the oxidative stress can be based upon endmember spectrum 2 and endmember spectrum 5 of the 8 endmember spectra. A decrease in the endmember spectrum 2 and an increase in the endmember spectrum 5 can be associated with an oxidative stress in the RBC membrane indicating the condition or predisposition for developing the condition. The control reference HSI spectra can be obtained from a red blood cell (RBC) reference spectral library. The method can comprise generating the RBC reference spectral library using hyperspectral dark field microscopy. The comparison can be based upon Spectral Angle Mapping (SAM).

[0003] In some aspects, the method can comprise positioning a blood sample of the subject on a microscope stage of the VNIR hyperspectral imaging system; identifying an acquisition area of the blood sample; and obtaining at least one HSI of the acquisition area. The identification of the acquisition area can be based at least in part upon shape and isolation of RBCs. The identification of the acquisition area can be further based upon lack of background noise. The blood sample can comprise an anticoagulant. The anticoagulantDocket: 220312-2030 can be EDTA. The blood sample can be stabilized by resting for a period of time after addition of the anticoagulant. The method can comprise obtaining a plurality of HSI of RBCs membranes in the acquisition area to obtain HSI spectra.

[0004] Other systems, methods, features, and advantages of the present disclosure will be or become apparent to one with skill in the art upon examination of the following drawings and detailed description. It is intended that all such additional systems, methods, features, and advantages be included within this description, be within the scope of the present disclosure, and be protected by the accompanying claims. In addition, all optional and preferred features and modifications of the described embodiments are usable in all aspects of the disclosure taught herein. Furthermore, the individual features of the dependent claims, as well as all optional and preferred features and modifications of the described embodiments are combinable and interchangeable with one another. BRIEF DESCRIPTION OF THE DRAWINGS

[0005] Many aspects of the present disclosure can be better understood with reference to the following drawings. The components in the drawings are not necessarily to scale, emphasis instead being placed upon clearly illustrating the principles of the present disclosure. Moreover, in the drawings, like reference numerals designate corresponding parts throughout the several views.

[0006] FIGS.1A-1D include examples of collected hyperspectral imaging (HSI) images and Spectral Angle Mapping (SAM) mapping results of four pure compounds, in accordance with various embodiments of the present disclosure.

[0007] FIGS.2A-2C illustrate examples of spectral libraries of the pure compounds, in accordance with various embodiments of the present disclosure.

[0008] FIGS.3A-3B and 4A-4B illustrate endmember spectra 2 and 5, in accordance with various embodiments of the present disclosure.Docket: 220312-2030

[0009] FIG.5 illustrates an example of a VNIR hyperspectral imaging system integrated into a microscope assembly, thereby enabling pixel-level spectral quantification of a sample being imaged, in accordance with various embodiments of the present disclosure.

[0010] FIG.6 illustrates an example of a workflow for the acquisition of hyperspectral images, in accordance with various embodiments of the present disclosure.

[0011] FIGS.7A-7D illustrate examples of red blood cell (RBC) spectral libraries, in accordance with various embodiments of the present disclosure.

[0012] FIG.8 includes examples of collected images and hyperspectral imaging results, in accordance with various embodiments of the present disclosure.

[0013] FIG.9 is a flow chart illustrating an example of the image acquisition and evaluation of spectral distribution, in accordance with various embodiments of the present disclosure.

[0014] FIGS.10A-10C include examples of collected images and hyperspectral imaging results, in accordance with various embodiments of the present disclosure.

[0015] FIGS.11A-11E include examples of collected HSI images and SAM mapping results of control and treated RBC samples, in accordance with various embodiments of the present disclosure.

[0016] FIG.12 illustrates an example of a distribution of RBC spectra of healthy subjects, in accordance with various embodiments of the present disclosure.

[0017] FIG.13 illustrates an example of RBC membrane analysis with HSI, in accordance with various embodiments of the present disclosure.

[0018] FIG.14 is a schematic block diagram of a computing device that can be used with the VNIR hyperspectral imaging system of FIG.5, in accordance with various embodiments of the present disclosure. DETAILED DESCRIPTION

[0019] Disclosed herein are various examples related to hyperspectral imaging (HSI) of red blood cell (RBC) membranes. Reference will now be made in detail to the description ofDocket: 220312-2030 the embodiments as illustrated in the drawings, wherein like reference numbers indicate like parts throughout the several views.

[0020] An RBC image can be examined by hyperspectral dark field microscopy (HDFM) (see FIG.5) with the steps illustrated in FIG.6, to obtain a spectral library (see examples of newly acquired in FIG.7A and former in FIG.7B) with specific bands (FIGS.7C and 7D, respectively). The distribution of 8 spectra in the optical image (see FIG.8) can be evaluated by class distribution (see FIG.9) and it is also possible to identify specific spectral libraries and bands due to molecular components used as references.

[0021] FIG.1A shows examples of collected HSI images and Spectral Angle Mapping (SAM) mapping results of lecithin liposomes from the analysis. FIG.1B shows examples of collected HSI images and SAM mapping results of POPC liposomes from the analysis. FIG. 1C shows examples of collected HSI images and SAM mapping results of cholesterol from the analysis. FIG.1D shows examples of collected HSI images and SAM mapping results of cholesterol oxide from the analysis.

[0022] Following image acquisition, by meticulously examining each image of the samples, a spectral library can be built for each one. In FIGS.1A-1D, representative images are presented for each compound accompanied by the resulting mapping after applying the library that was built for each one. FIGS.2A-2C illustrate examples of the libraries. Along with the four libraries shown in FIG.2A, a detailed list of the bands for each spectral endmember is provided in FIG.2B. Lecithin liposomes, POPC and cholesterol need 5 spectra to be fully described, while cholesterol oxide needs only 4. The Savitzky-Golay filtering method was applied to smooth the original signal data. The construction of the spectral library allows the detection of each individual band of the spectral endmembers and specifically attribute selected blood library bands to pure compounds library bands, thus attributing specific spectra bands to specific molecular groups.

[0023] For the attribution of the bands, it is only reasonable to work with the strongest of the peaks. In FIG.2C, the strong bands of each of the 8 spectral endmembers of RBCDocket: 220312-2030 spectral library are characterized based on the homogeneity with the pure compound peaks. For each characterized band an indication arrow is given showing whether the RBC is shifted to the left (↓), to the right (↑), or it is stable (-).

[0024] The use of the reference material spectra can be superimposed with the RBC spectral library to get similitudes (see FIGS.3A-3B and FIGS.4A-4B). In FIG.3A, the two spectra of main interest -spectra 2 and 5- are placed in the same graph with the libraries of each compound. Each library endmember is given in the same color (red), while spectra 2 and 5 are given in green and cyan color as usual. From the table in FIG.2B, the two spectra of main interest, phospholipids, are primarily affecting. Certainly, the band attribution to the pure compounds is better overviewed after the normalization of the curves. In FIG.3B, the graphs of FIG.3A are normalized to the unity.

[0025] FIGS.4A and 4B illustrate enlarged images of (a) POPC spectrum 2, cholesterol spectrum 2 and RBC spectrum 2, and (b) POPC spectrum 2, Lecithin spectrum 2 and RBC spectrum 5, respectively. In each case, the strong peaks are noted for the erythrocyte and the pure compound along with an arrow indicating the orientation of the shift going from the compound to the expanded system of an erythrocyte. In the same graphs, the RBC strong peaks are noted along with the peaks of pure compounds spectra. Additionally, for each peak, an arrow is placed showing the orientation of the shift going from the fundamental system of the pure compound to the complex system of a red blood cell. As far as the RBC spectrum 2 is concerned, a blue shift occurs for the peaks at 518 nm and 559 nm, since POPC’s peaks decrease from 519 nm and 560 nm, respectively. On the contrary, for the band at 593 nm a red shift is observed with phospholipid’s peak increasing by 2 nm. A fourth band at 648 nm is common for both systems (RBC and cholesterol oxide). Interestingly, for spectrum 5 there is no blue shift since the bands at 520 nm, 562 nm and 594 nm are all red shifted by 1-2 nm. An extra peak at 608 nm remains constant both for POPC liposomes and erythrocytes.

[0026] Mainly, the use of the specific molecular components, such as phospholipids, lecithins, cholesterol, or cholesteryl oxide as reference compounds and the identification ofDocket: 220312-2030 their hyperspectral properties are important in order to identify the composition of the sample assembly, in particular to know that the sample is “membrane” and not other materials.

[0027] This disclosure deals with the possibility of detection and measurement of the organizational features of the cell membrane at a nanoscale level, using HDFM and considering the arrangement of the membrane molecular cluster as a whole object (nano-bio material). This assembly in vivo is known to be subjected to changes / reactions due to “environmental” cellular conditions, however the HDFM tool was never utilized to observe the “RBC object” before and after exposure to agents. When illuminated by light and examined by hyperspectral imaging techniques, a new nano-detection of membrane assembly properties, as it were a “surface,” is demonstrated.

[0028] In applications to health, HDFM of RBC samples can be used with RBC to compare healthy and unhealthy patients. The possibility of using the RBC spectral library as “descriptor” of RBC changes due to the reactivity of the “surface” has never been explored. In this disclosure, erythrocyte membranes are considered to contain information per se: after having established the RBC spectral library, that identifies this “material”, it can be used to consequently identify what happens when this “object” is exposed to specific conditions that alter its assembly. The methodology concerns the most common exposure in a biological environment, i.e., the so-called OXIDATIVE STRESS which can modify cells at various extent, with or without destroying them, and can be also associated with unhealthy conditions. Indeed, oxidative stress is a physio-pathological condition, which has been linked to many diseases [Forman, H.J., Zhang, H. Targeting oxidative stress in disease: promise and limitations of antioxidant therapy. Nat Rev Drug Discov 20, 689–709 (2021)], and therefore it is a frequent condition that cell (and in our case RBC) encounters in the biological environment.

[0029] RBCs are ideal samples to be examined under such conditions for at least two reasons:1) as “material”, they are known to be very resistant to physical and chemical stress, able to maintain their morphology and the information embedded therein; 2) as biological information, their molecular composition and asset allow normal functions, asDocket: 220312-2030 oxygen / nutrient transporter and general cellular homeostasis, therefore any insult to the assembly can represent an information also from the biological / clinical point of view.

[0030] HSI technology was first applied for remote sensing in space and agricultural applications, and then moved to light microscopy detection. HSI is a non-invasive technique that uses visible light and a camera to capture a very accurate digital image description, measuring the dispersed incoming radiation into certain wavelengths in each pixel, thus acquiring hundreds of pixels in a “HS cube”. Through the analysis of its spectral and spatial features, the “object” is characterized by its signatures or fingerprints and can be recognized also in its variations. Applications in dark and bright field microscopy gave the opportunity to examine a variety of samples, like tissues, cells, molecules, nanomaterials, both ex-vivo or in vivo, by label methodology.

[0031] Absorption-based hyperspectral imaging was used in “Absorption-Based Hyperspectral Imaging and Analysis of Single Erythrocytes” by Ji Youn Lee et al. (IEEE Journal of Selected Topics in Quantum Electronics 18.3 (2011), pp.1130–1139) to resolve unique physicochemical characteristics of subcellular substances in single erythrocytes is reported. By constructing a hyperspectral imaging microscope system installed with a spectral light engine capable of controlling the spectral shape of the illumination light by a digital micromirror device, and sequentially using maximum angle convex cone algorithm, unique spectral signatures (i.e., endmembers) for three different types of hemoglobin (oxyhemoglobin, methemoglobin, and hemozoin) and scatter from cell membrane in single erythrocytes. Additionally, further statistical endmember analysis on the hyperspectral image data was realized, thus providing the abundances of specific endmembers, while a modeling based on Mie scattering theory to explain the scattering signatures as a function of scattering angle was also realized. The developed imaging and analysis technique enables label-free molecular imaging of endogenous biomarkers in single erythrocytes in order to build oximetric standards on a cellular level and ultimately for in vivo as well.

[0032] Hyperspectral dark field microscopy (HDFM) has been used for probing and characterizing cell membrane images of red and white blood cells. A HDFM approach canDocket: 220312-2030 be used to create a human RBC reference spectral library that can image and afford a new comprehensive descriptor of the RBC. The library made of 8-endmembers was mapped on the image using with a spectral angle mapper algorithm, and a percentual distribution of the 8 endmembers was obtained in the RBC images. In this study molecular components typical of cell membranes, such as phospholipids [i.e., lecithins] and cholesterol, as well as of hemoglobin [i.e., protoporphyrin] were used to recognize specific spectral characteristics of such components in the whole spectral signatures of the RBC membrane. The RBC spectral library was applied in the study of Giacometti et al. to compare the HDFM RBC images in children affected by Autism Spectrum Disorder (ASD) and healthy age-matched subjects, showing a statistically significant difference (mainly in one of the 8-membered spectral library) in their distribution maps. This result preliminarily showed the possibility of detection of the RBC membrane impairment involved in diseases using HDFM methodology.

[0033] The RBC can be used as the representative cell, focusing on the cell membrane, for the identification human body status. The RBC molecular asset has been correlated with health status in membrane lipidomics research, and in autism, obesity, and cancer. The molecular basis for application of the HDFM technology to RBC membranes allows the use of spectral library detection with the identification of different spectral distributions to characterize the specific molecular phenomenon which affects membranes.

[0034] Described herein is the use of a visible near infrared (VNIR) hyperspectral imaging system as shown in FIG.5 as a non-invasive diagnostic tool for the analysis of RBC membrane oxidative status. Also described herein is the use of a VNIR hyperspectral imaging system in high throughput screening of oxidative stress in RBC membranes. In certain aspects, an important principle is the use of hyperspectral imaging (HSI) as a technique that integrates conventional imaging and spectrophotometry and enables pixel level spectral quantification of the sample being imaged. As an example, the Cytovia-HSI system used captures the VNIR (400-1000 nm) spectrum within each pixel of the scanned field of view with superior signal to noise ratio. By evaluating an expanded spectrum ofDocket: 220312-2030 transmitted light well beyond the visible spectrum, additional information is gained to further characterize the cellular changes caused by oxidative stress.

[0035] This technology can be used for early detection of conditions such as, e.g., autism in humans as well. This method can serve as a tool for high throughput screening. While existing technologies may provide detection of autism at very later stages, the present methodology provides very early detection based upon oxidative stress in the red blood cell (RBC) membrane. Furthermore, this methodology is non-invasive and does not require administration of external dyes / reagents for detection. A HSI scanning technique can be employed for screening of an individual. The changes induced by a condition of a patient may be detected and confirmed. Furthermore, the HSI scan for early detection even before observable changes may occur.

[0036] RBC detection by hyperspectral imaging allows for identification of the membrane assembly and characterization of the spectral distribution of the library in the optical image, as described in the case of autism. Healthy and unhealthy patients have been compared, in order to identify differences in the hyperspectral imaging, using the distribution scores of the 8 spectra that form the library of the human RBC. Here it is demonstrated that a permanent impairment of the membrane asset can be determined by oxidative stress and can be detected and measured by HDFM. Oxidative stress has an impact on the membrane assembly by causing oxidation of polyunsaturated lipids (see, e.g., “Membrane changes under oxidative stress: the impact of oxidized lipids” by R. Itri et al. Biophys Rev.2014 Mar; 6(1):47-61), also activating membrane turnover that is very important to keep health and longevity (see, e.g., “Membrane phospholipids, lipoxidative damage and molecular integrity: A causal role in aging and longevity” by R. Pamplona. Biochem. Biophys. Acta 2008, 1777, 1249). Therefore, oxidative stress cannot be considered only destructive, but it can have regenerative effects.

[0037] The HDFM tool and the image processing can examine with a high precision the changes in the membrane asset in order to individuate membrane perturbation, and establish its thresholds, eventually linked to a disease condition or to be used in a preventiveDocket: 220312-2030 manner. HDFM is a very sensitive detection technique – easy to use from a micro-blood drop, without additional probes – and it is here developed to establish the impact of oxidative stress on RBC membrane. H2O2is considered mainly responsible of oxidative stress in the body, activating signaling for antioxidant response (see, e.g., “Resistance to H2O2-induced oxidative stress in human cells of different phenotypes” by V. Zenin et al. Redox Biology, 2022,50,102245), but it was never used to evaluate changes in the RBC hyperspectral features.

[0038] Certain embodiments provide a method for determining whether a subject has a condition or is predisposed for developing the condition, comprising obtaining a hyperspectral image (HSI) of a RBC membrane over a range of wavelengths to obtain a HSI spectrum and determining whether the spectrum is indicative of oxidative stress, wherein the presence of the oxidative stress indicates the individual has a condition such as, e.g., autism or is predisposed for developing the condition.

[0039] In certain embodiments, the wavelengths used to obtain the HSI spectrum can be in the visible near infrared (VNIR) range. The HSI spectrum can be compared to at least a first previous HSI spectrum obtained from the subject at an earlier point in time, wherein significant spectral difference between the HSI spectra indicates the subject has the condition or is predisposed for developing the condition. The HSI from the subject can be compared to a control reference HSI spectrum and to a condition reference HSI spectrum to determine whether the image comprises spectral differences that are indicative of the condition.

[0040] FIG.6 illustrates an example of a workflow for the acquisition of hyperspectral images. Initially, a blood sample is extracted from an individual and the whole blood treated in EDTA collected from healthy human subjects. The EDTA is the anticoagulant. A sample is then loaded on a slide and allowed to stabilize. For example, 2 μL of the whole blood sample can be placed on the microscope glass slide and sandwiched with a coverslip. This procedure can be repeated three times (preparation of three different slides) and the 3 slides left to rest and stabilize for about one hour. The number of samples (n=3) can be chosen toDocket: 220312-2030 provide a sufficient number of repeatable measurements (3 RBC for each slide, n=9 RBC) to evaluate the measurement deviation and errors.

[0041] After stabilization, the slide can be placed on the microscope stage for evaluation. Each of the prepared glass slides can be positioned on the microscope’s stage and the focus along z-axis, as well as on x-y plane set. An important aspect of the experimental process is the location of a proper acquisition area for the image acquisition. The appropriate area should be chosen to contain several nicely round shaped and well- illuminated red blood cells. The RBCs should be isolated from other RBCs to avoid cell-cell interaction effects. Additionally, it is beneficial for the selected areas to be free of background noise (e.g., noise caused by various nanoparticles or plasma). With the slide in proper position, image acquisition can begin.

[0042] As soon as the proper area has been located, the exposure time can be adjusted on a value which sets the maximum intensity, e.g., at about 5000 counts at 640 nm. (the intensity level used on the experiment was 10x higher than the level used in previous works, thus decreasing SNR noise and increasing the levels of confidence). At this point, 1’- 2’ of waiting was needed to allow the photoluminosity of the cells to decrease. This time is needed to avoid that acquired reflectance spectra be covered by a “shoulder”-like behavior in blue-green wavelength area values, and the resulting images are deteriorated.

[0043] Once the area is chosen and exposure time has been set-up, The image can be taken in the first slide after the area is chosen and the exposure time set-up, and then the analysis of the image is performed. A plurality of RBCs (e.g., three) can be chosen in the image to analyze the spectral mapping and the distribution of the spectral end-members. The imaging and distribution mapping can then be repeated for the other two slides. For each blood sample, a number of RBC images (e.g., 9) are finally obtained and examined. Time of detection is not distant from the time of blood collection. Generally, the detection occurs within 24 hours from the blood withdrawal.

[0044] Dark-field images can be recorded at room temperature using an enhanced dark-field illumination system (CytoViva, Auburn, AL) attached to an Olympus microscopeDocket: 220312-2030 (EDFM). The system can comprise a CytoViva 150 dark field condenser in place of the microscope original condenser attached via a fiber optic light guide to the lamp source. A 60x oil immersion colour corrected objective can be integral to the system. A 150 W quartz halogen light source (Dolan Jenner DC-950, Massachusetts, USA) can be used, which covers the full spectrum from 400 nm to 2500 nm. The images can be collected on a range from 400 nm to 1000 nm (VNIR).

[0045] The system employs a darkfield-based illuminator that focuses a highly collimated light at oblique angles on the sample to obtain images with improved contrast and signal-to-noise ratio. A concentric imaging spectrophotometer containing aberration- corrected convex holographic diffraction gratings can be used with 1.29 nm spectral resolution (12.5 μm slit). The utilization of this concept can obviate the necessity of extraneous agents (e.g., staining or contrast agents), as the core imaging technique relies on light scattering / transmittance and resolution of scattered / transmitted light in the visible and near-IR wavelengths. Light scattering is a property unique to every type of insoluble particulate matter, ergo, spectral information relayed by this technique correlates directly to the location and type (or size) of particulate matter in question. ENvironment for Visualization (ENVI ver.4.4) software can be employed to control the resolution of scattered light and aggregation of the subject of interest.

[0046] The images can be captured at a defined magnification using the ENVI software, which is followed by extraction of spectral curves. The camera can capture spectral responses from 400-1000 nm with a maximum readout time of 12 frames per second. To aid in the identification of oxidative stress in unknown samples, a spectral library can be created as will be discussed. After establishment of this reference library, the spectral angle mapping algorithm can be employed for comparison of the distinct spectral features of the library with the spectral scan of the unknown sample.

[0047] The CytoViva Hyperspectral image analysis software powered by ENVI contains features to aid in the quantification and identification of materials by spectral data analysis. At the beginning of each hyperspectral image analysis, images can be corrected for darkDocket: 220312-2030 current values obtained after blocking illumination. Radiation emanating from illuminated cells can be collected on the objective lens, in turn translated into an image on the entrance- slit plane, which can in turn be projected onto the prism-grating-prism components. Consequently, the radiation changes its direction of propagation depending on the wavelength. Each area of the cells can be represented as a set of monochromatic points on the detector, translating continually into a spectrum along the spectral axis. Movement of the objective lens amounts to variation in the region captured along with variation the wavelength captured. A set of such images captured at varying camera locations thus yielded all regions and wavelengths of the sample.

[0048] For the spectral library, 8 spectral endmembers with distinct characteristics and configuration are recognized in the RBC image. Differences of the present procedure may be attributed to the better set-up of the exposure time, with a clear amelioration of the background and sample noise. This change of the detection procedure led to a repetition of the acquisition of the RBC spectral library. FIGS.7A and 7B show examples of an actual library and a previous library, respectively, to evidence the differences of the two acquisitions, where the Savitzky-Golay filtering method was applied to smooth the original signal data. In FIGS.7A and 7B, the y-axis emphasizes the different intensity order these two libraries. The main spectral bands of the 8 spectral endmembers are collected in the table of FIG.7C for the new detection and the table of FIG.7D for the previously reported detection. In FIG.7C, the strongest peaks found in each of the 8 endmembers in the new acquisition are shown and, in FIG.7D, the peaks found in each of the 8 endmembers in the old acquisition are shown. The spectra numbers (spectrum 1,2,3,4,5,6,7,8) are arbitrary and do not correspond between the new and old numbering. Focusing on the specific bands, one can see that many of them appear in both tables (e.g., 560 nm, 520 nm, 595 nm), while in the new library some extra bands appear too (e.g., 545 nm).

[0049] Among supervised classification methods, a Spectral Angle Mapping (SAM) classification algorithm was selected due to its better performance than other classification algorithms for detection of oxidative stress in unknown samples. The SAM algorithm is a toolDocket: 220312-2030 for rapid mapping of the spectral similarity of image spectra to reference spectra. The SAM is a physically based spectral classification that uses an n-dimensional angle to match pixels to reference spectra. This method determines the spectral similarity by calculating the angle between the spectra, treating them as vectors in a space with dimensionality equal to the number of bands or wavelengths. The angle between the endmember spectrum vector and each pixel vector in n-dimensional space is then compared. Smaller angles represent closer match to a reference spectrum. Pixels further away from the specified maximum angle threshold in radians are not classified.

[0050] By collecting the number of the peaks in each image pixel, SAM algorithm superimposes reflectance spectra in an n-D space as vectors, where n is the number of peaks for each spectrum and classifies each pixel using a threshold angle between each pixel vector and each spectral endmember. The smaller the angle between a pixel and an endmember, the closer is the match between them. By assigning a color code to the spectral endmembers identified in the sample, the optical image can be converted to a false-colored spectral imaging, utilizing many color channels, and the satisfactory match between optical and spectral data can be visualized. In case that a pixel is unable to match at an angle smaller than the threshold with any of the endmembers, it can be considered unclassified, and the pixel can be colored black. FIG.8 illustrates an example of an optical image and the SAM mapping in a representative RBC sample. The color code assigned to the spectral endmember allows to see that the optical image is fully covered by the library.

[0051] Analysis of Cell Data. For every collected image, individual cells of interest can be taken into account for the analysis. Using the ROI, the distribution of the 8 spectral endmembers in the image pixels can be given as distribution percentages and the resulting data of each cell image can be normalized to the number of classified pixels (i.e., reject all unclassified, calculate the sum of the classified and report each spectral distribution relative to this sum). By using this normalization, the size of each cell can be taken into consideration.Docket: 220312-2030

[0052] FIG.9 shows a flow chart illustrating the image acquisition, as illustrated in FIG. 8, and evaluation of spectral distribution. After the images are acquired, the SAM classification algorithm processes the images as discussed and single cells can be selected as the regions of interest (ROI) for analysis. The selected cells can then be analyzed to determine the spectral endmember distribution and data acquisition.

[0053] RBC are exposed to a plethora of endogenous oxidant species in vivo, and the oxidation of hemoglobin in RBC with formation of oxyhemoglobin (HbO2) is one of the most important processes to monitor, such as for example during infections. Hydrogen peroxide (H2O2) is endogenously produced during by immune-system and endothelial cells for various metabolic processes, therefore it is reasonable to use this as a chemical reagent to expose RBC to oxidative conditions. The use of hyperspectral imaging to monitor the oxidative processes occurring to RBC has never occurred. In particular, the method regarding the monitoring of the lipid components by HDFM is new. The method is not limited to H2O2but can include other biological, chemical and physical conditions affecting cells and cell membranes.

[0054] Choosing the H2O2concentration for experiments, the concentration was set at 3%, considering two limits: a lower limit with no efficiency below 0.5%, and an upper limit at 10% with cell destruction. The treatment procedure for a sample size of 50 μL EDTA-whole blood included adding 3% of H2O2and leaving it in a glass vial at room temperature for 1 hour. After the incubation, the blood sample was centrifuged (5000 rpm at 4 °C) and the supernatant (water, plasma and remaining of H2O2) was removed leaving only the packed RBC.2 μL of this sample was then put as a drop on the glass slide and sandwiched with the cover slip. The procedure was repeated 3 times and each time 3 RBC images were chosen to obtain a sufficient number of images (n=9) for statistical analysis.

[0055] FIG.10A shows an example of a typical image collected on an oxidized sample with its corresponding mapping after SAM algorithm is applied. A single red blood cell found in the HSI result is used for the analysis, while in the corresponding mapping of the RBC is shown. It is noted that each mapping color corresponds to the colors of the 8 endmembersDocket: 220312-2030 as depicted in the library demonstrated in FIG.7A. Using the cell images in FIG.10A and the flowchart of the image processing in FIG.9, the data on the RBC exposed to H2O2are shown in FIG.10B. The detailed statistical data are shown, from top to bottom, for each of the three H2O2(3%) RBC sets as well as the total values, calculated from each set’s total values. FIG.10C shows a comparison of the HDFM spectral distribution between a healthy RBC and the H2O2(3%)-treated RBC.

[0056] The class distribution data are given in graph form in FIG.10D, with the delta variations from the healthy to oxidized RBCs. Each dot represents the average value with its standard error of the three replicates (controls and oxidized are represented), while shaded areas represent total average values with the corresponding errors. The δ-variations for each spectrum are also noted. Regarding the analysis, even though many of the spectra result in interesting variations and seem promising, it was decided to focus mainly on the statistical data of spectrum 2 and spectrum 5 as they comprise two spectra with representative statistical behavior.

[0057] Spectrum 2 shows a strong behavior for the control samples with the average value located at 15.05 % and the standard error at 1.30%, while the oxidized samples indicate a significant drop in spectrum 2’s presence since the average value plummets to 10.10% and the standard error is about the same (1.28%). With the δ-variation at 4.95 %, or at -33% of the relevant ratio between the two, it can be seen that there is a significant drop of 1 / 3 in the expected value of spectrum 2’s presence on the RBC membrane as soon as the hydrogen peroxide is added.

[0058] Spectrum 5 is the most abundant among the 8 endmembers and -in contrast with spectrum 2- it is increased in the oxidized samples than in healthy RBC, reaching 24.30% during the presence of H2O2 and 17.81% in controls. As far as the standard errors are concerned, healthy RBC show a smooth behavior at 1.10% while in the presence of H2O2the standard error drops impressively at 0.06%. These two expected values suggest a δ-variation of 6.49%, or an increase by 36% of spectrum 5’s presence after the oxidative reaction.Docket: 220312-2030

[0059] For pure compounds analysis, two RBC organic component groups were taken into account: phospholipids and cholesterol. A thorough analysis was possible by preparing egg lecithin liposomes, POPC, cholesterol and cholesterol oxide samples. After fixing the concentration value for each sample, microscope glass slides were prepared and the image acquisition was realized using the same procedure as RBCs. Suspensions in PBS of cholesterol and cholesterol oxide powders were prepared at 4.2 mM, Liposome samples were also submitted to centrifugation and subsequent PBS washing in order to remove all excess background interference. For the microscope slide preparation, approximately 5 μL were used for cholesterol and cholesterol oxide, and 10 μL for the liposomes.

[0060] The model used as reference for oxidative stress measurement is the RBC oxidation with 3% hydrogen peroxide (corresponding to 0.88M) for 1 hour at room temperature. Under these conditions RBCs do not change morphology (see FIGS.11A and 11B for comparison, with the RBC image before and after oxidation), are not destroyed, but they receive a consistent insult to the membrane assembly. In FIGS.11A-11C, the images and the distributions of the RBC spectral library before and after the experiments (repeated with 3 different blood samples in triplicates and using 3 different RBCs from each sample,^^ ൌ 9) are shown. FIG. 11A illustrates (a) a typical hyperspectral image (HSI) collected on acontrol sample with (b) its corresponding mapping after SAM algorithm is applied. FIG.11A shows (c) a single red blood cell found in the HSI (a) that is used for the analysis, and (d) the corresponding mapping of the cell. It is noted that each mapping color corresponds to the colors of the 8 endmembers as depicted in the library demonstrated in FIG.7A. FIG.11B illustrates (a) A typical HSI collected on an oxidized sample with (b) its corresponding mapping after the SAM algorithm is applied. FIG.11B shows (c) a single red blood cell found in the HSI (a) that is used for the analysis, and (d) the corresponding mapping of the cell. It is noted that each mapping color corresponds to the colors of the 8 endmembers as depicted in the library demonstrated in FIG.7A.

[0061] FIG.11C illustrates the spectral distributions of the RBC spectral library in the images acquired from RBC samples before and after exposure to 3% H2O2for 1 hour atDocket: 220312-2030 room temperature. The data are mean ± SEM of three samples repeated three times for each sample and for 3 RBC in each sample. FIG.11D shows a summary of the distribution of the 8 RBC spectra before and after H2O2oxidative experiments, indicating the variations (positive and negative) of each value. The variations recorded in the oxidative stress samples are compared to the distribution found in a RBC sample taken from a child affected from Autism Spectrum Disorder (ASD).87.5% of the spectra are useful to identify the variation of the EXAMPLE RBC ASD similar to the RBC oxidation. FIG.11E graphically illustrates the distribution of the 8 RBC spectra before and after H2O2oxidative experiments, indicating the variations (positive and negative) of each value, reported in FIG.6D.

[0062] FIG.12 shows the comparison of the RBC spectral library distribution in twogroups of healthy subjects, differing from the age (children (^^ ൌ 20) vs. adults (^^ ൌ 10), andthe delta values and statistics show that Spectra 1 and 6 have a significant variation. The summary of the distribution of the 8 RBC spectra in healthy children (n=20) and healthy adults (n=10) is shown on the top and the variations (delta) are shown and the statistical significance of these variations has been evaluated in the table shown below evidencing only spectrum 1 and 6 with significant differences.

[0063] From these experiments, by extensive use of this methodology, it will be possible to evidence more and more what are the best spectra to take for diagnostic development. In the data presented for this application, spectrum 2 and spectrum 5 are envisaged as the most interesting for the follow-up of oxidative conditions, being the most affected by the H2O2treatment (FIG.6D, Delta = -5.35 for spectrum 2 and Delta = 5.09 for spectrum 5) whereas both spectra present the smallest differences for their distributions between adults and children (FIG.7, 0.45 and 0.29).However, none of the data should be dismissed for their diagnostic values.

[0064] EXAMPLE: The proposed method of HDFM spectral distribution in RBC membrane can describe a differential number for various oxidative stress conditions (which is comprehensive of different membrane rearrangements, simulating health conditions). It is foreseen that the extent of this number is linked to the specific pathology. DIFFERENTIALDocket: 220312-2030 between the spectral distributions (Delta positive and negative) are calculated in FIG.11D from the control and oxidative experiment and compared with a RBC sample from a patient affected with autism spectrum disorder (ASD). It can be seen that 7 out of the 8 spectra follow the same trend (positive and negative) found in the oxidative experiment.

[0065] Specific changes in PL containing PUFA: Note that the oxidative experiments were coupled with the analysis of the membrane fatty acid composition, determining the changes of fatty acids, in particular polyunsaturated fatty acids (PUFA) which are the most sensitive to the oxidative stress conditions. The results of the oxidized vs control samples of RBC showed the fatty acid distribution represented in FIG.13, which illustrates membrane fatty acid analysis of the RBC controls and treated with 3% H2O2, with the HDFM data shown in FIG.11D. The significant changes of the fatty acids occur for the diminution of polyunsaturated fatty acids (PUFA) and the increase of saturated fatty acids (SFA). These molecular changes are expected to connect with the changes of the membrane asset detected by the HDFM spectral distribution.

[0066] With reference now to FIG.14, shown is a schematic block diagram of a computing device 1400 that can be used with the VNIR hyperspectral imaging system of FIG.5 according to an embodiment of the present disclosure. The computing device 1400 includes at least one processor circuit, for example, having a processor 1403 and a memory 1406, both of which are coupled to a local interface 1409. To this end, the computing device 1400 may comprise, for example, at least one server computer or like device. The local interface 1409 may comprise, for example, a data bus with an accompanying address / control bus or other bus structure as can be appreciated.

[0067] Stored in the memory 1406 are both data and several components that are executable by the processor 1403. In particular, stored in the memory 1406 and executable by the processor 1403 are an image acquisition application 1412, a data analysis application 1415, and potentially other applications 1418. The image acquisition application 1412 and / or the data analysis application 1415 can implement, when executed by the computing device 1400, various aspects of the computational processing as described above withDocket: 220312-2030 respect to the workflow / flowchart of FIGS.6 and 9. For example, the image acquisition application 1412 can facilitate acquisition and / or storage of acquired hyperspectral and optical images and the data analysis application 1415 can facilitate processing of the HSI data. In some implementations, the image acquisition application 1412 and data analysis application 1415 may be combined in a single application. Also stored in the memory 1406 may be a data store 1421 including, e.g., recordings, images, video, HSI spectra and other data. In addition, an operating system may be stored in the memory 1406 and executable by the processor 1403. It is understood that there may be other applications that are stored in the memory and are executable by the processor 1403 as can be appreciated.

[0068] A hyperspectral imaging (HSI) device 1424 (e.g., a spectrophotometer such as a Specim V10E Imspector spectrograph, a Headwall spectrograph, etc. as shown in FIG.5) and an optical imaging device 1427 (e.g., an optical camera such as a PCO Panda 4.2 color camera, a Dage optical camera, etc. as shown in FIG.5) in communication with the computing device 1400 can be utilized to obtain optical images and their corresponding HSI data (e.g., an HSI spectra of the RBC membrane). Synchronization of each optical image and its HSI data can be coordinated by the image acquisition application 1412. For example, the image acquisition application 1412 can provide a control signal to initiate acquisition of the optical image and its HSI data to one or both of the HSI and optical imaging devices 1424 and 1427. The acquired optical image and corresponding HSI data can be stored in memory by the image acquisition application 1412 with a time stamp and / or frame number. After capturing and storing the data, the data analysis application 1415 can produce the movie or series of frames including the optical and HSI information.

[0069] Where any component discussed herein is implemented in the form of software, any one of a number of programming languages may be employed such as, for example, C, C++, C#, Objective C, Java®, JavaScript®, Perl, PHP, Visual Basic®, Python®, Ruby, Delphi®, Flash®, or other programming languages. A number of software components are stored in the memory and are executable by the processor 1403. In this respect, the term "executable" means a program file that is in a form that can ultimately be run by theDocket: 220312-2030 processor 1403. Examples of executable programs may be, for example, a compiled program that can be translated into machine code in a format that can be loaded into a random access portion of the memory 1406 and run by the processor 1403, source code that may be expressed in proper format such as object code that is capable of being loaded into a random access portion of the memory 1406 and executed by the processor 1403, or source code that may be interpreted by another executable program to generate instructions in a random access portion of the memory 1406 to be executed by the processor 1403, etc. An executable program may be stored in any portion or component of the memory including, for example, random access memory (RAM), read-only memory (ROM), hard drive, solid- state drive, USB flash drive, memory card, optical disc such as compact disc (CD) or digital versatile disc (DVD), floppy disk, magnetic tape, or other memory components.

[0070] The memory 1406 is defined herein as including both volatile and nonvolatile memory and data storage components. Volatile components are those that do not retain data values upon loss of power. Nonvolatile components are those that retain data upon a loss of power. Thus, the memory 1406 may comprise, for example, random access memory (RAM), read-only memory (ROM), hard disk drives, solid-state drives, USB flash drives, memory cards accessed via a memory card reader, floppy disks accessed via an associated floppy disk drive, optical discs accessed via an optical disc drive, magnetic tapes accessed via an appropriate tape drive, and / or other memory components, or a combination of any two or more of these memory components. In addition, the RAM may comprise, for example, static random access memory (SRAM), dynamic random access memory (DRAM), or magnetic random access memory (MRAM) and other such devices. The ROM may comprise, for example, a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), an electrically erasable programmable read- only memory (EEPROM), or other like memory device.

[0071] Also, the processor 1403 may represent multiple processors 1403 and the memory 1406 may represent multiple memories 1406 that operate in parallel processing circuits, respectively. In such a case, the local interface 1409 may be an appropriateDocket: 220312-2030 network that facilitates communication between any two of the multiple processors 1403, between any processor 1403 and any of the memories 1406, or between any two of the memories 1406, etc. The processor 1403 may be of electrical or of some other available construction.

[0072] Although portions of the image acquisition application 1412, data analysis application 1415, and other various systems described herein may be embodied in software or code executed by general purpose hardware, as an alternative the same may also be embodied in dedicated hardware or a combination of software / general purpose hardware and dedicated hardware. If embodied in dedicated hardware, each can be implemented as a circuit or state machine that employs any one of or a combination of a number of technologies. These technologies may include, but are not limited to, discrete logic circuits having logic gates for implementing various logic functions upon an application of one or more data signals, application specific integrated circuits having appropriate logic gates, or other components, etc. Such technologies are generally well known by those skilled in the art and, consequently, are not described in detail herein.

[0073] The image acquisition application 1412 and data analysis application 1415 can comprise program instructions to implement logical function(s) and / or operations of the system. The program instructions may be embodied in the form of source code that comprises human-readable statements written in a programming language or machine code that comprises numerical instructions recognizable by a suitable execution system such as a processor in a computer system or other system. The machine code may be converted from the source code, etc. If embodied in hardware, each block may represent a circuit or a number of interconnected circuits to implement the specified logical function(s).

[0074] Although the workflow / flowchart of FIGS.6 and 9 shows a specific order of execution, it is understood that the order of execution may differ from that which is depicted. For example, the order of execution of two or more blocks may be scrambled relative to the order shown. Also, two or more blocks shown in succession in FIGS.6 and 9 may be executed concurrently or with partial concurrence. Further, in some embodiments, one orDocket: 220312-2030 more of the blocks shown in FIGS.6 and 9 may be skipped or omitted (in favor, e.g., measured travel times). In addition, any number of counters, state variables, warning semaphores, or messages might be added to the logical flow described herein, for purposes of enhanced utility, accounting, performance measurement, or providing troubleshooting aids, etc. It is understood that all such variations are within the scope of the present disclosure.

[0075] Also, any logic or application described herein, including the image acquisition application 1412 and data analysis application 1415 that comprises software or code can be embodied in any non-transitory computer-readable medium for use by or in connection with an instruction execution system such as, for example, a processor 1403 in a computer system or other system. In this sense, the logic may comprise, for example, statements including instructions and declarations that can be fetched from the computer-readable medium and executed by the instruction execution system. In the context of the present disclosure, a "computer-readable medium" can be any medium that can contain, store, or maintain the logic or application described herein for use by or in connection with the instruction execution system.

[0076] The computer-readable medium can comprise any one of many physical media such as, for example, magnetic, optical, or semiconductor media. More specific examples of a suitable computer-readable medium would include, but are not limited to, magnetic tapes, magnetic floppy diskettes, magnetic hard drives, memory cards, solid-state drives, USB flash drives, or optical discs. Also, the computer-readable medium may be a random access memory (RAM) including, for example, static random access memory (SRAM) and dynamic random access memory (DRAM), or magnetic random access memory (MRAM). In addition, the computer-readable medium may be a read-only memory (ROM), a programmable read- only memory (PROM), an erasable programmable read-only memory (EPROM), an electrically erasable programmable read-only memory (EEPROM), or other type of memory device.Docket: 220312-2030

[0077] It should be emphasized that the above-described embodiments of the present disclosure are merely possible examples of implementations set forth for a clear understanding of the principles of the disclosure. Many variations and modifications may be made to the above-described embodiment(s) without departing substantially from the spirit and principles of the disclosure. All such modifications and variations are intended to be included herein within the scope of this disclosure and protected by the following claims.

[0078] The term "substantially" is meant to permit deviations from the descriptive term that don't negatively impact the intended purpose. Descriptive terms are implicitly understood to be modified by the word substantially, even if the term is not explicitly modified by the word substantially.

[0079] It should be noted that ratios, concentrations, amounts, and other numerical data may be expressed herein in a range format. It is to be understood that such a range format is used for convenience and brevity, and thus, should be interpreted in a flexible manner to include not only the numerical values explicitly recited as the limits of the range, but also to include all the individual numerical values or sub-ranges encompassed within that range as if each numerical value and sub-range is explicitly recited. To illustrate, a concentration range of “about 0.1% to about 5%” should be interpreted to include not only the explicitly recited concentration of about 0.1 wt% to about 5 wt%, but also include individual concentrations (e.g., 1%, 2%, 3%, and 4%) and the sub-ranges (e.g., 0.5%, 1.1%, 2.2%, 3.3%, and 4.4%) within the indicated range. The term “about” can include traditional rounding according to significant figures of numerical values. In addition, the phrase “about ‘x’ to ‘y’” includes “about ‘x’ to about ‘y’”.

Claims

Docket: 220312-2030 CLAIMS Therefore, at least the following is claimed:

1. A method for determining whether a subject has a condition or is predisposed for developing the condition, comprising using a visible near infrared (VNIR) hyperspectral imaging system to obtain a hyperspectral image (HSI) of a red blood cell (RBC) membrane of the subject over a range of wavelengths to obtain HSI spectra and determining whether the HSI spectra is indicative of oxidative stress in the RBC membrane, wherein the presence of the oxidative stress in the RBC membrane indicates the subject has the condition or is predisposed for developing the condition, wherein the wavelengths used to obtain the HSI spectra are in the VNIR range, wherein the method is non-invasive and does not comprise administration of dyes for detection of the oxidative stress in the RBC membrane.

2. The method of claim 1, wherein the condition is autism.

3. The method of any of claims 1 and 2, wherein the HSI spectra of the RBC membrane is compared to control reference HSI spectra to determine delta variations characterizing a status of the RBC membrane, the status associated with the oxidative stress of the RBC membrane.

4. The method of claim 3, wherein the HSI spectra comprise 8 endmember spectra of the RBC membrane.

5. The method of claim 4, wherein the delta variations comprise increases or decreases in defined wavelength ranges in the endmember spectra.Docket: 220312-2030 6. The method of claim 4, wherein the status of the RBC membrane associated with the oxidative stress is based upon endmember spectrum 2 and endmember spectrum 5 of the 8 endmember spectra.

7. The method of claim 6, wherein a decrease in the endmember spectrum 2 and an increase in the endmember spectrum 5 is associated with an oxidative stress in the RBC membrane indicating the condition or predisposition for developing the condition.

8. The method of claim 3, wherein the control reference HSI spectra are obtained from a red blood cell (RBC) reference spectral library.

9. The method of claim 8, comprising generating the RBC reference spectral library using hyperspectral dark field microscopy.

10. The method of claim 3, wherein the comparison is based upon Spectral Angle Mapping (SAM).

11. The method of claim 1, comprising: positioning a blood sample of the subject on a microscope stage of the VNIR hyperspectral imaging system; identifying an acquisition area of the blood sample; and obtaining at least one HSI of the acquisition area.

12. The method of claim 11, wherein the identification of the acquisition area is based at least in part upon shape and isolation of RBCs.Docket: 220312-2030 13. The method of claim 12, wherein the identification of the acquisition area is further based upon lack of background noise.

14. The method of claim 11, wherein the blood sample comprises an anticoagulant.

15. The method of claim 14, wherein the anticoagulant is EDTA.

16. The method of claim 14, wherein the blood sample is stabilized by resting for a period of time after addition of the anticoagulant.

17. The method of claim 11, comprising obtaining a plurality of HSI of RBCs membranes in the acquisition area to obtain HSI spectra.

Images