Diagnosis and localization of disease state using quantitative parameter array set of feature fusion

Abstract

The universal biomarker array enables diagnosis and localization of disease states earlier than now recognizable, and overall increases the accuracy of diagnosing disease states. A universal biomarker array is formed from a series of mathematical parameters calculated from digital tissue images showing tissue details at cellular levels. The universal biomarker array includes one or more novel arrays of parameters (i.e., a spatial entropy array, a bin array, and / or a quartile array) in combination with one or more arrays of prior art parameters (i.e., a pattern array, a distance array, and / or a morphological array) and / or in combination with one or more other parameters. The AI / ML system can be used to analyze universal biomarker arrays relative to database-based reference data for early diagnosis and localization of disease processes.

Description

[0001] Cross-reference to related applications

[0002] This patent application claims the benefit of U.S. Provisional Patent Application No. 63 / 464,058, filed May 4, 2023, entitled UNIVERSAL BIOMARKER ARRAY, which is incorporated herein by reference in its entirety. Technical Field

[0003] This invention relates in general to a universal biomarker array that can be used across multiple tissue types and diseases, combined with AI / machine learning / statistical analysis, for early disease diagnosis, disease process localization, and real-time feedback, and can be used during clinical and scientific imaging and image analysis. Background Technology

[0004] In clinical medicine and basic science, the relevant focus is on improving methods for diagnosing the early disease process (early diagnosis) of many diseases, thereby making more accurate diagnoses and precisely locating early disease stages or diseases within tissues, preferably in conjunction with near-immediate feedback.

[0005] Currently, depending on the type of tissue and disease, there is a range of specialized existing diagnostic methods for non-clinical and clinical use, such as laboratory (blood, urine tests), functional and genetic tests, clinical imaging (magnetic resonance imaging (MRI), CT scans), endoscopy and taking biopsies, such as for histology or RNA-Seq. Non-clinical tests (e.g. basic science assays) can well determine and localize early disease processes, but often require tissue biopsies and tissue-destructive analysis, which is either impossible or ethically unjustifiable for all tissues. Clinical tests (e.g. genetic, laboratory and functional tests) do not allow localization of disease processes; they determine the risk for a certain disease, the systemic presence of disease-specific biomarkers or the restricted extent of disease caused. Liquid biopsies are also non-location specific. In contrast, imaging tests are important for localizing disease, but they are limited by their technical resolution. In some studies, MRI scanners achieve a spatial resolution of less than 0.5 mm, while clinical MRI scanners achieve a spatial resolution of about 1-2 mm in routine imaging studies, but often do not reach the resolution to clinically visualize early disease processes, e.g. on a microscopic scale. Thus, MRI is not sensitive enough to detect small changes in tissue structure at early stages of disease. Modern endoscopy (e.g. confocal microscopic endoscopy) can have microscopic resolution and can be used to identify early disease in near real-time using disease-specific markers and downstream methods described herein, in order to accurately diagnose early disease. Other uses of images of organs / tissues / tissue sections for medical diagnosis include evaluation by a pathologist or other expert relying on professional experience or by an AI / machine learning system relying on existing technology parameters. SUMMARY

[0006] According to one embodiment of the present application, a method and system for disease diagnosis comprises receiving a digital tissue image showing tissue details at cellular level; performing automatic image analysis on the received digital tissue image to identify cells of interest for analysis; calculating a series of mathematical parameters at multiple levels to collectively serve as a universal biomarker array, the universal biomarker array comprising a pattern array, a distance array, a morphology array, a spatial entropy array, a bin array where it is converted to relative information by assigning absolute parameter values to specific bin locations, and a quartile array where absolute parameter values are converted to relative information by assigning absolute parameter values to specific quartiles; and analyzing the universal biomarker array relative to database-based reference data for early diagnosis and localization of disease processes.

[0007] In various alternative implementations, the digital tissue image can comprise a microscopic endomicroscopy digital tissue image. Performing automatic image processing can involve identifying cells of interest relative to an image background, constructing at least one region of interest (ROI) in non-background regions, and performing ROI-based segmentation to (i) segment cells for analysis of the segmented cells and (ii) analyze inverse segmentation for analysis of spaces between the cells. The array of spatial entropy parameters can comprise Batty (absolute, relative), Contagion, Karlstrom (absolute, relative), O Neill (absolute, relative), and Parredw parameters. Computing the array of bins can involve assigning each of the plurality of parameters into a relative class and a range-specific bin location within an entire range of parameter values, wherein a total number of bins is computed by multiplying a number of disease states by a number of bins per state, and wherein the conversion of data into relative class-specific and range-specific bin locations can be performed for parameters of the array of spatial entropy, the array of patterns, the array of distances, and the array of morphology. Computing the array of quartiles can involve assigning each of the plurality of parameter values to a disease state-specific quartile location and converting each of the plurality of parameter values to a disease range-specific quartile location, wherein the conversion of data into relative class-specific and range-specific quartile locations can be performed for parameters of the array of spatial entropy, the array of patterns, the array of distances, and the array of morphology. Analyzing the universal biomarker array relative to database-based reference data can involve providing the universal biomarker array to an AI / ML system trained on universal biomarker array data to detect early diagnosis and localization of disease processes, wherein the AI / ML system can utilize random forest regression to detect early diagnosis and localization of disease processes.

[0008] Additional implementations can be disclosed and claimed. BRIEF DESCRIPTION OF DRAWINGS

[0009] The advantages of various implementations of the present application will be more fully understood in view of the following "Description of Illustrative

[0010] Figure 1 A random forest classification of KIN vs. SCC using a universal biomarker array is shown according to certain implementations.

[0011] Figure 2 A predictive classification of colon adenoma vs. cancer using a universal biomarker array is shown according to certain implementations.

[0012] Figure 3FIG. 1 is a graph showing the significant increase in classification accuracy of human joint cartilage disease states due to the use of novel parameter arrays (n=3) and novel unique array combinations as different predictive modeling input datasets (n=57) compared to prior art parameters (n=3).

[0013] Figure 4 FIG. 1 is a graph showing the significant increase in classification accuracy of human joint cartilage disease states due to the use of novel parameter arrays (n=3) and novel unique array combinations as different predictive modeling input datasets (n=57) compared to prior art parameters (n=3).

[0014] Figure 5 FIG. 1 is a graph showing the significant increase in classification accuracy of human joint cartilage disease states due to the use of novel parameter arrays (n=3) and novel unique array combinations as different predictive modeling input datasets (n=57) compared to prior art parameters (n=3).

[0015] Figure 6 FIG. 1 is a graph showing the significant increase in classification accuracy of human joint cartilage disease states due to the use of novel parameter arrays (n=3) and novel unique array combinations as different predictive modeling input datasets (n=57) compared to prior art parameters (n=3).

[0016] Figure 7 FIG. 1 is a graph showing the significant increase in classification accuracy of human joint cartilage disease states due to the use of novel parameter arrays (n=3) and novel unique array combinations as different predictive modeling input datasets (n=57) compared to prior art parameters (n=3).

[0017] Figure 8 FIG. 1 is a graph showing the significant increase in classification accuracy of human joint cartilage disease states due to the use of novel parameter arrays (n=3) and novel unique array combinations as different predictive modeling input datasets (n=57) compared to prior art parameters (n=3).

[0018] Figure 9 FIG. 1 is a graph showing the significant increase in classification accuracy of human joint cartilage disease states due to the use of novel parameter arrays (n=3) and novel unique array combinations as different predictive modeling input datasets (n=57) compared to prior art parameters (n=3).

[0019] It should be noted that the foregoing figures and elements depicted therein are not necessarily drawn to scale or any particular scale. Identical elements are denoted by identical numbers, unless the context clearly dictates otherwise. The figures are primarily for illustrative purposes and are not intended to limit the scope of the inventive subject matter described herein. DETAILED DESCRIPTION

[0020] Definitions. Unless the context requires otherwise, the following terms have the indicated meanings as used in the specification and claims:

[0021] A“set” includes one or more members, even if the set description is presented in the plural (e.g., a set of multiple Xs can include one or more Xs).

[0022] Depending on the context, the concept of“real-time” can range from instantaneous to seconds or minutes. For example, real-time feedback for a medical procedure would typically require providing feedback during the course of the medical procedure, and in some cases can require providing feedback within a desired time frame (e.g., if tissue damage will occur within X minutes of detecting the start of a disease state / process such as loss of oxygen, then real-time feedback needs to be provided within X minutes for it to be useful in taking remedial action).

[0023] The term“disease” as in disease state, process, diagnosis, etc. is used herein generally to refer to an attribute or combination of attributes indicative of some distinct condition, which can be a disease, an injury, a defect, an allergy or chemical / drug reaction, over-activity, under-activity, healing, regeneration, growth, or other condition that can be analyzed using the techniques described herein.

[0024] In view of the above background, it is clear that there is a need to improve the sensitivity, accuracy, localization, and speed of feedback of methods of detecting / diagnosing and localizing early disease processes. Three short examples highlight the advantages of methods that can achieve the above points:

[0025] • Early detection of cancer A method that can non-invasively and in near real-time detect early stages of certain skin or other tumors would immediately become clinically relevant, as early treatment would be within reach,

[0026] • Early detection of tissue damage / derangement With current clinical techniques, we cannot determine early tissue damage / denaturation, which is an early, potentially reversible disease state of osteoarthritis (OA), a degenerative joint disease that remains a major cause of disability worldwide. Thus, we cannot test new / promising candidate drugs early enough in the early disease state that can be sensitive to treatment, thus missing the opportunity. A method that can detect early stages of OA (or other diseases) would allow the development of new, disease state-specific drugs and / or testing of existing, promising drugs that fail in clinical trials because they are tested too late in the disease process,

[0027] • Real-time feedback: Feedback on whether a given tissue region is affected by a disease would improve the local accuracy of diagnostic biopsies (skin, intestine), tumor resections (resection margins), and precise implantation of regenerative medicine implants during surgery (treatment of cartilage defects).

[0028] Overall, the ability to diagnose and localize earlier, more accurately, combined with real-time feedback, will significantly improve the development of treatment protocols and treatment outcomes and will yield opportunities to meaningfully improve medical pathways.

[0029] To be able to diagnose and localize disease states earlier than is currently possible, and to improve the accuracy of diagnosing disease states in general, the inventors have developed a collection of the best performing quantitative parameter arrays through feature fusion. Specifically, certain embodiments compute various mathematical parameters (e.g., spatial entropy, cell morphology, distribution and density, and population-based parameters as described / defined below) from images of organs, tissues, or tissue sections (generally referred to herein as “tissue”) that show details at the cellular level, which are collectively used to quantitatively describe the health of the organ / tissue or tissue section and specific disease states across tissues and disease types. For convenience, the collection of relevant mathematical parameters can be logically categorized into separate arrays (where each separate array essentially translates different aspects of healthy tissue architecture and changes in diseased tissue architecture into comprehensive quantitative parameters based on images taken with cellular resolution / method), and multiple arrays can be collectively analyzed as a feature fusion collection (referred to herein as “universal biomarker arrays”). It should be noted that the term “universal” here reflects that the biomarker arrays are applicable to various organs / tissues or tissue sections and / or health and disease states, and that various embodiments can use different arrays and / or arrays with different constituent mathematical parameters as universal biomarker arrays, e.g., for different organs / tissues and / or for detecting different health and disease states. The universal biomarker arrays can be provided as predictive modeling input data to artificial intelligence / machine learning systems, e.g., for classifying health and disease states or for basic statistics.

[0030] Furthermore, such quantitative biomarker arrays and related methodologies can be used, e.g., for testing images / data of interest against database-based reference data specific for health and / or disease states, supported by statistics and / or artificial intelligence (AI) / machine learning (ML), such as:

[0031] • Distinguishing between health and early disease states in organs / tissues or tissue sections (early disease diagnosis)

[0032] • Localization of disease processes in organs / tissues or tissue sections

[0033] • Real-time feedback on disease states and localization during imaging in organs / tissues (decision support tools, e.g., during surgical procedures)

[0034] • Monitoring disease processes and tissue architecture (e.g., healing / regeneration, drug treatment outcomes, cosmetic application outcomes, etc.)

[0035] • Diagnosis and staging of tumors

[0036] • Improving diagnosis of disease state in organs / tissues or tissue sections by improving accuracy compared to the state of the art

[0037] It should be noted that while the term “early” detection is often referred to herein, the technology described herein can provide benefits in other applications. For example, later stage disease diagnosis can benefit from such a universal biomarker array as it can be better than current methods.

[0038] Certain embodiments innovatively quantitatively describe healthy and early disease tissue architecture with mathematical parameters (called a universal biomarker array) that describe cell population characteristics and matrix characteristics, which are used as AI / ML inputs for diagnosis or other statistical analysis, and can be used whenever imaging / images allow identification / segmentation of cells (and matrix) within an organ / tissue / tissue section and imaging / images of the cells.

[0039] In general, the cell population characteristics of a given tissue are significantly tissue type specific and also disease sensitive, e.g., cells are lost in trauma, cells are more spaced apart in tissue swelling, and cell numbers increase, cell locations are very different, or different types of cells are present in cancer and other proliferative diseases, compared to healthy tissue. Thus, a series of mathematical parameters that quantify a range of characteristics of the tissue cell population together constitute a disease sensitive digital fingerprint of tissue architecture and function. Similarly, the extracellular matrix of a given tissue - the space between cells - can undergo various changes in early disease, which can be described with mathematical parameters in healthy and early disease. These data are used as AI inputs for early diagnosis and disease localization across tissue types and diseases, thus constituting a universally available biomarker array for diagnosis.

[0040] The AI / ML inputs used here are different from other diagnostic AI / ML applications that use as input: (i) annotated medical images without further quantification, e.g., for convolutional neural network analysis, or (ii) simpler quantification using segmented images, e.g., Haralick texture features, or (iii) where segmentation does not identify cell populations / matrix, e.g., because images from MRI (or other imaging modalities) are at too low resolution to analyze cells / matrix.

[0041] The following is a description of the workflow according to certain embodiments.

[0042] The workflow typically starts by generating images of the organ / tissue / tissue section (e.g., with endomicroscopy, microscopy, or other imaging techniques) and / or using existing images of the organ / tissue / tissue section (e.g., from histological sections generated for evaluation by a pathologist or other expert). Importantly, as noted above, embodiments of the present application operate on images with resolution that depict cells. A particular method for generating images for the analyses described herein is to use commercially available endomicroscopes for clinical use, e.g., from the company Mauna Kea Technologies (e.g., Cellvizio, a probe-based confocal laser endomicroscopy system) or from the company Zeiss (e.g., CONVIVO, a confocal endomicroscopy system) or other suitable systems for clinical use. A particular method for using images from histological sections for the analyses described herein is to use a digital pathology slide scanner to generate the images or to use existing scans. A particular method for generating non-clinical images from an organ / tissue / tissue section for the analyses described herein is to use any type of microscope, e.g., fluorescence, confocal, or other microscopy techniques. If the images are generated for clinical use (e.g., using endomicroscopy), certain embodiments can use an auto-fluorescent drug, e.g., using an antibiotic such as tetracycline or ciprofloxacin, or using quinine as a fluorescent dye, or a fluorescent molecule (i.e., where such molecule is not a drug in a pharmacological sense), as there are currently no clinically available fluorescent cell dyes available. It should be noted that the use of drugs (e.g., antibiotics) is not here because of their pharmacological activity. This is not necessary for all tissues except some tissues (e.g., articular cartilage, where cells are not easily recognized). If the auto-fluorescent drug does not easily penetrate the tissue (e.g., cartilage), then drug polarization is performed, e.g., by adjusting the Ph / using another charged / polar molecule in the solution. For example, tetracycline penetrates into negatively charged tissues for visualizing cells of the tissue using tetracycline as a fluorescent dye, e.g., in cartilage, can be achieved by adding a positive charge to the solution to interact with the negatively charged tetracycline functional group. This can be achieved with polar solvents like PEG and glucose solutions (e.g., 70% or 20%). In particular, a tetracycline solution for clinical use (e.g., a doxycycline vial for intravenous injection) can be used in conjunction with a solution of about 20-70% glucose for clinical use. If needed, tissue penetration of ciprofloxacin and other auto-fluorescent polar drugs can be achieved by similar means. Since doxycycline can interact with divalent cations like magnesium and calcium, resulting in increased fluorescence, divalent cations can be bound to doxycycline for imaging, by adding e.g. magnesium ions, or by using isotonic solutions that are commonly used for arthroscopic surgery that contain divalent cations like magnesium ions. In this case, the fluorescent signal can be excited e.g. with UV-A light in the range of 350-400 nm.Another possibility of using autofluorescent drugs as fluorescent dyes to visualize tissue cells is to synthesize drug derivatives with positively charged groups (e.g. amino groups) attached to them. It should be noted that other imaging techniques (including future developed imaging techniques or even enhanced / improved versions of current imaging techniques that do not have cellular resolution (e.g. existing MRI and CT imaging techniques)) can be used in alternative embodiments of the present invention.

[0043] The workflow typically continues with automatic image analysis for identifying organs / tissues / tissue sections / cells in the image from the image background, constructing regions of interest (ROIs) in non-background areas, and segmentation based on ROIs for (i) segmenting cells for analyzing the segmented cells and (ii) analyzing inverse segmentation for analyzing the space between the cells.

[0044] The workflow typically continues with computing a series of mathematical parameters at multiple levels, which collectively serve as a universal biomarker array, and are described / defined below.

[0045] The workflow typically continues with storing the curated (e.g. organ / tissue / tissue section / cell and disease state specific) diagnostic / classification / annotation universal biomarker array data in a database to produce database based reference data.

[0046] The workflow typically continues with using the image / data of interest for statistical and / or AI / ML supported testing against the database based reference data for early diagnosis and localization of disease processes. One particular AI method is using random forest classification for diagnostic purposes, e.g. testing (i) whether “new” data from an image of interest represents a healthy or diseased state, and / or (ii) determining a specific disease classification / score. Here, “new” data from a “new” image of interest is tested against reference data, which can be stored in a database or otherwise available to the system. Another particular AI method is using random forest regression to compute a continuous numerical value, e.g. a continuous score or other value of interest, where “new” data from a “new” image of interest is tested against database reference data. Another particular AI method is using LightGBM classification as described for random forest classification. Another particular AI method is using PLS-DA for discriminant analysis (e.g. with max.dist, centroids.dist, and / or mahalanobis.dist) and / or using sPLS-DA to select a subset of variables for discriminant analysis and prediction / classification of “new” data. Typically, the accuracy, precision, recall, and F1 score of each model are used to decide on a particular AI model.

[0047] Thus, the workflow determines whether the image / data of interest is representative of healthy organ / tissue / tissue section data or deviates from healthy data, e.g., a specific disease state can be diagnosed and localized, and also the accuracy of the diagnosis (classification) and other quality control related parameters can be determined. Moreover, because the embodiments allow for analysis of tissue and tissue sections, the universal biomarker array is applicable to cells within tissue and cells visible in tissue sections. Thus, for example, the universal biomarker array can be particularly applicable to pathology institutes (tissue sections) and clinical trials / routine (patient / live cell imaging). Moreover, because the embodiments allow for analysis using different staining solutions (dyes), but also allow for analysis using live cell imaging, the universal biomarker array of the present invention is not dependent on dyes.

[0048] The following is a description of various mathematical parameters of the universal biomarker array currently contemplated for use with certain embodiments, which can be used as AI / ML input or other statistical analysis for early diagnosis and disease localization across tissue types and diseases:

[0049] • Calculated using the "R" language and library "SpatEntropy" from the point pattern created from the cell coordinates:

[0050] • Spatial Entropy Batty (batty)

[0051] • Spatial Entropy Contag (contag)

[0052] • Spatial Entropy Karlstrom (karlstrom)

[0053] • Spatial Entropy O'Neill's (oneill)

[0054] • Spatial Entropy Parredw (parredw)

[0055] • Spatial Entropy Leibovici (leibovici)

[0056] • Spatial Entropy Altieri (altieri),

[0057] • Absolute and relative values used when available

[0058] • Calculated using the "R" language and library "SpatEntropy" from the labeled point pattern created from the cell coordinates, using all parameters calculated for each cell as labels:

[0059] • Spatial Entropy Shannon (shannon) using absolute and relative values

[0060] • Calculated using the "R" language and library "spatstat":

[0061] o For each cell:

[0062] • Nearest neighbor distance (nndist)

[0063] o For each image ROI / image of the point pattern created from the cell coordinates:

[0064] • Mean and average nndist

[0065] • 1. Quartiles of nndist

[0066] • 2. Quartiles of nndist

[0067] • 3. Quartiles of nndist

[0068] • Standard deviation of nndist

[0069] • Standard error of nndist

[0070] • Spatial intensity

[0071] • Clark-Evans aggregation index (clarkevans) calculated without edge correction, "cdf" edge correction, and "Donnelly" edge correction

[0072] • Mean and average of the Clark-Evans aggregation index version

[0073] • Pair correlation function maximum (pcfmax) calculated with "trans" and "iso" correction

[0074] • r-value of pair correlation function maximum on y-axis (x-axis) (calculated with Ohser-Stoyan translation and Ripley isotropy correction)

[0075] • Calculated using ImageJ / Fiji or "R" language:

[0076] o For cells:

[0077] • Length

[0078] • Width

[0079] • Aspect ratio

[0080] • Roundness

[0081] • Circularness

[0082] • Solidity

[0083] • Perimeter

[0084] • Area

[0085] • Direction

[0086] • For each image ROI / image / parameter, if applicable:

[0087] • Mean, median, standard deviation, standard error, quartiles

[0088] The above listed parameters focus on the cell population of a given tissue and are computed using cell recognition / segmentation. The inverse of the segmentation mask selects the extracellular matrix, i.e. the space between cells within the tissue. From the segmentation / area data, quantitative parameters can be computed, such as the spatial entropy parameters (see above for parameters; library "SpatEntropy") and the Haralick texture features.

[0089] Certain embodiments introduce three novel arrays of quantitative parameters:

[0090] (a) A novel array of quantitative parameters, called spatial entropy (abbreviated as Entropy), whose various spatial entropy parameters are computed from the cell positions (details below).

[0091] (b) A novel array of quantitative parameters, called bins, where absolute parameter values are converted into relative information by assigning them to specific bin positions. Binning refers to the process of dividing a continuous set of parameter data into consecutively numbered intervals (bins) and assigning each data point to a specific bin number according to its value. Specifically, each parameter value is converted into a disease state-specific bin position, using 20 bins per disease state to determine separately for each disease state. In addition, each parameter value is converted into a disease range-specific bin position. Here, bin positions across the entire range of parameter values are assigned, and the total number of bins is computed by multiplying the number of disease states by the number of bins per state (e.g. 20 bins). So far, the conversion of data into relative categories and range-specific bin positions is performed on the novel entropy array and the prior art arrays of morphology, patterns, and distances, but in the future it is not necessarily limited to these.

[0092] (c) A novel array of quantitative parameters, called quartiles, where absolute parameter values are converted into relative information by assigning them to specific quartile positions. Specifically, each parameter is assigned to a disease state-specific quartile position, and the computation is performed independently for each state. In addition, each parameter value is converted into a disease range-specific quartile position, where the entire range of values is used for the quartile computation and each value is assigned to these quartiles. So far, the conversion of data into relative category-specific and range-specific quartile positions is performed on the entropy, morphology, patterns, and distances arrays, but in the future it is not necessarily limited to these.

[0093] In one exemplary embodiment, the entropy array includes the spatial entropy parameters Batty (absolute, relative), Contagion, Karlstrom (absolute, relative), O'Neill (absolute, relative), and Parredw, which can be computed with the R package "SpatEntropy." It is important to note that the exact number and / or type of spatial entropy parameters presently contemplated for the entropy array should not be understood as an exclusive final list, as the number and / or type of spatial entropy parameters can vary in different embodiments. In this regard, it is important to note that the exact number and / or type of parameters presently contemplated for any array should not be understood as an exclusive final list, as the number and / or type of parameters can vary in different embodiments.

[0094] In certain embodiments, the entropy, box, and quartile arrays are used in combination with one or more of the following additional arrays (sometimes referred to as prior art arrays) as a feature fusion set that can be used for predictive modeling:

[0095] (d) a pattern array, which in one embodiment includes the following parameters: spatial intensity, Clark-Evans index without edge correction, Clark-Evans index with cdf edge correction, Clark-Evans index with Donnelly edge correction, pair correlation function maximum value and x-value at which maximum is observed (with both Ohser-Stoyan translation and Ripley isotropy correction), which can be computed with the R package "spatstat."

[0096] (e) a distance array, which in one embodiment utilizes individual nearest neighbor distances determined for each cell and its nearest neighbor (e.g., using the spatstat package for R), and in one embodiment includes computed mean, median, and standard deviation, as well as first, second, and third quartiles.

[0097] (f) a morphology array, which in one embodiment utilizes individual cell morphology determined for each cell, and in one embodiment includes computed mean values for cell area, length, width, orientation, perimeter, circularity, and solidity.

[0098] It should be noted that while certain embodiments can use all six of these arrays and optionally also use other parameters, such as Figure 9 as illustratively depicted in FIG. 2, alternative embodiments can include one or more novel arrays in combination with one or more prior art arrays and / or one or more novel arrays in combination with one or more other parameters. Furthermore, as noted above, the constituent parameters of each array can vary between different embodiments.

[0099] To generate different unique array combinations and compare the disease state classification accuracy of individual arrays and unique array combinations with existing state-of-the-art parameters as predictive modeling inputs, a series of morphological, pattern, distance, entropy, box, and quartile array values were calculated for a range of diseases / disease states in three human tissues: articular cartilage, skin, and colon (details below). The arrays were used as individual arrays (a total of 6; 3 are state-of-the-art: pattern, distance, morphology) and to generate unique array combinations (combinations of 2, 3, 4, 5, and 6 arrays; a total of 57 unique combinations). This resulted in 63 different arrays / array combinations that were used as different predictive modeling inputs for random forest modeling. The modeling process in “R” included 5-fold cross-validation and 5-fold cross-validation and was performed twice, using a balanced class (disease state) distribution, with an equal number of data rows for each class for all runs. This step allowed the analysis of the resulting accuracy as a function of the 63 different predictive modeling inputs. Figures 3 to 5 Statistical analysis was performed to identify those individual arrays and unique array combinations with significantly higher classification accuracy than the state-of-the-art parameters. Statistical tests were performed on three data subsets. Each included all novel arrays, all novel unique array combinations, and one of the three state-of-the-art parameter sets as detailed in the legends of Figures 3 to 5 .

[0100] In human articular cartilage ( Figure 3 ), statistical tests showed that the disease state classification accuracy of all novel arrays and all unique array combinations were significantly different from the accuracy produced using the state-of-the-art parameter arrays: pattern (p < 0.001), distance (p < 0.001), and morphology (p < 0.001). As shown in Figure 3 , in articular cartilage, the use of novel array boxes alone and the use of most unique array combinations as predictive modeling inputs (except for the entropy pattern, distance entropy, distance pattern, distance entropy pattern array combinations) resulted in significantly higher classification accuracy than using the state-of-the-art parameters, clearly outperforming the state-of-the-art parameters.

[0101] Figure 3 is a plot showing the significant increase in classification accuracy of human articular cartilage disease states due to the use of novel parameter arrays (n = 3) and novel unique array combinations as different predictive modeling input datasets (n = 57) compared to the state-of-the-art parameters (n = 3) as detailed in Figure 3As shown, box plots illustrate the overall accuracy resulting from using different predictive modeling input datasets; they give the median and the 25th and 75th percentiles, and whisker lines give the 10th and 90th percentiles. Line plots illustrate the specific accuracy calculated for each disease state (category) using the confusion matrix generated by modeling. Individual disease states describe how cells are arranged in spatial patterns (e.g., in strings, double strings, small clusters, and large clusters, as well as diffuse arrangements without discernible spatial patterns), as such arrangements are associated with structural and functional pathology (details described below). These disease state-specific accuracies indicate the accuracy of correctly classifying a particular disease state and indicate how well the model performs for a single disease state. The Shapiro-Wilk test for the normality of the test data shows that all datasets are non-normally distributed. The Wilcoxon rank-sum test, performed in pairs with p-values ​​adjusted using the Bonferroni method, shows that each prior art array pattern, distance, and morphology is significantly different from each novel array and each unique array combination (each p < 0.0005). Therefore, no significant differences were shown.

[0102] In human skin ( Figure 4 Statistical tests showed that the disease state classification accuracy of all novel arrays and all unique array combinations was significantly different from the accuracy produced by using existing techniques with parametric array patterns (p<0.001), distance (p<0.001), and morphology (p<0.001) as predictive modeling inputs. Figure 4 As shown, compared with using existing technical parameters, located in Figure 4 Different array combinations on the right side of the morphological array on the x-axis produce significantly higher classification accuracy in human skin, thus significantly exceeding the performance achieved using current technical parameters.

[0103] Figure 4 The graph illustrates the significant increase in classification accuracy for human skin disease states compared to existing technical parameters (n=3), resulting from the use of novel parameter arrays (n=3) and novel, unique array combinations as different predictive modeling input datasets (n=57). Figure 4The box plots show the overall accuracy resulting from using different predictive modeling input datasets; they give the median as well as the 25th and 75th percentile, and the whiskers give the 10th and 90th percentile. The line plots illustrate the specific accuracy calculated for each disease state (class) using the confusion matrix produced by the modeling. The individual disease states are actinic keratosis with a grading of KIN (intraepithelial neoplasm of keratinocytes) I, II, or III, which is a premalignant lesion, and moderately differentiated squamous cell carcinoma (SCC G2), which is a malignant cancer (described in detail below). These disease state specific accuracies indicate the accuracy of correctly classifying a specific disease state and indicate how well the model performs for the individual disease states. The Shapiro-Wilk test for normality of the test data shows that all data sets are non-normally distributed. The Wilcoxon rank sum test with pairwise comparisons using Bonferroni adjusted p-values shows that all novel arrays are significantly different in pattern, distance, and morphology (0.000001 < p < 0.007). In addition, each unique array combination is significantly different from the prior art and the novel individual arrays (0.000001 < p < 0.007). Therefore, no significant differences are shown.

[0104] In the human colon (Figure 2), Figure 5 Statistical tests revealed multiple significant differences in classification accuracy when using the prior art technique versus the novel array entropy and a series of unique array combinations, which are detailed in Figure 2. As Figure 5 shown, the series of unique array combinations (indicated with "§") resulted in significantly higher classification accuracy in the human colon than using the prior art parameters as predictive modeling input, clearly outperforming the prior art parameters.

[0105] Figure 5 is a plot showing the significant increase in classification accuracy of human colon disease states due to using the novel parameter arrays (n=3) and the novel unique array combinations as different predictive modeling input datasets (n=57) compared to the prior art parameters (n=3), as Figure 5The box plots show the overall accuracy resulting from using different predictive modeling input datasets; they give the median as well as the 25th and 75th percentiles, and the whiskers give the 10th and 90th percentiles. The line plots illustrate the specific accuracy calculated for each disease state (class) using the confusion matrix produced by the modeling. The individual disease states are colon adenoma and colon carcinoma (see text for details on the tissues, diseases, and disease states analyzed). These disease state specific accuracies indicate the accuracy of correctly classifying a specific disease state, and indicate how well the model performed for the individual disease state. The Shapiro-Wilk test for normality of the test data shows that all datasets are non-normally distributed. The Wilcoxon rank sum test with pairwise comparisons using Bonferroni adjusted p-values shows that the novel entropy array is significantly different from each of the prior art array patterns, distances, and morphologies (each p < 0.0001), while the novel box and quartile arrays are not. In addition, the prior art array patterns and distances are significantly different from each unique array combination (0.000001 < p < 0.02). The prior art array morphologies are significantly different from a series of unique array combinations denoted by "m" (0.000001 < p < 0.02). No other significant differences are noted. The symbol "§" indicates unique array combinations that, when used as predictive modeling input, result in significantly higher classification accuracy than any of the prior art parameters.

[0106] Next, by pooling all tissues / disease states, the novel individual arrays and unique array combinations were tested against the prior art arrays as predictive modeling inputs in a complex environment, which resulted in 12 different classification choices. Using the pooled data (see text for details on the tissues, diseases, and disease states analyzed), Figure 6 statistical testing showed that all disease state classification accuracies reached significance (0.0000001 < p < 0.03) except for the comparison between array distance and entropy. Importantly, the novel array box and all different array combinations to the right of the morphological array on the x-axis located at Figure 6 produced significantly higher classification accuracies than the prior art parameters. Note that this superior performance was achieved in the context of cross-classification of different tissues and disease states, highlighting the efficacy of the novel, unique array combinations as predictive modeling inputs for classification tasks. This achievement demonstrates the practical utility of these unique array combinations for complex real-world scenarios, and affirms their potential as versatile tools for advanced classification applications. Thus, Figures 3 to 5 presents the disease state classification accuracies in specific tissues, Figure 6 presents the disease state classification accuracies in the pooled complex dataset. These Figure 1It was statistically convincingly demonstrated that a series of unique array combinations significantly outperformed the accuracy produced by the state-of-the-art parameters when used as predictive modeling inputs.

[0107] Figure 6 is a plot showing the significantly increased classification accuracy of a broad range of unique array combinations acting as different predictive modeling input datasets in the complex environment generated by pooling all tissue / disease states, as Figure 6 indicated, Figures 3 to 5 The individual tissue / disease states used in the are pooled for generating the challenging classification task. Box plots (pink) illustrate the overall accuracy produced using the different predictive modeling input datasets; they give the median as well as the 25th and 75th percentile, the whiskers give the 10th and 90th percentile. Line plots illustrate the specific accuracy calculated for each disease state (class) using the confusion matrix produced by the modeling. These disease state-specific accuracies indicate the accuracy of correctly classifying a specific disease state and indicate how well the model performs for a single disease state. The Shapiro-Wilk test for normality of the data shows that all datasets are non-normally distributed. The Wilcoxon rank-sum test using the Bonferroni method for pairwise comparison of p-values shows that the disease state classification accuracy of all novel arrays and all unique array combinations significantly differs from the accuracy produced by the state-of-the-art parameter arrays mode, distance, and shape used as predictive modeling inputs (0.0000001 < p < 0.03). The only exception is the comparison between array distance and entropy, which did not reach significance.

[0108] Next, the unique array combinations that consistently show top performance across all tissues, disease states, and the pooled complex dataset were identified. This was achieved by ranking each classification accuracy plot ( Figures 3 to 6 ) in the box plot by increasing median accuracy and, for each represented disease state, by increasing the line plot accuracy. Each ranking iteration documented the resulting order of unique array combinations, which captures their (ascending) performance as predictive modeling inputs for a specific disease state (data not shown). This approach led to the identification of the top 11 unique array combinations that consistently achieve the highest classification accuracy across all analyzed tissue / disease states. Figure 7 is a plot showing the state-of-the-art and novel single arrays and top unique array combinations that consistently achieve the highest classification accuracy across all analyzed tissue / disease states. The classification accuracy of the three state-of-the-art arrays (mode, distance, shape) and their combined array (mode_distance_shape) is shown relative to the top 11 unique array combinations in their respective tissue / disease states. Figure 8is a graph showing the best performing unique array combination of feature fusion sets and their average disease state classification accuracy performance across all tissue / disease states relative to the state-of-the-art parameters. The state-of-the-art parameters, the novel array, and the best performing feature fusion set of quantitative parameter arrays that "translate" different aspects of healthy tissue architecture and changes in diseased tissue architecture into comprehensive mathematical parameters as predictive modeling input data for classifying disease states are shown, significantly outperforming the accuracy of the state-of-the-art parameters.

[0109] In the above described embodiments, the array values are calculated using thresholded images of human articular cartilage tissue depicting a range of cells below the surface of the tissue and images of biopsy tissue sections of human skin and colon depicting a range of cells within the tissue.

[0110] According to publication [1], articular cartilage images depicting Calcein AM stained cells are classified according to how the cells are arranged in spatial patterns, e.g., strings, double strings, small clusters, and large clusters, and diffuse arrangements without discernible spatial patterns. These spatial patterns are indicative of structural [2] and functional pathology [8, 10]. Note that the content of these publications partially quantifies the state-of-the-art parameters distance and / or pattern. This is not in conflict with the present application, as the novel array and the best performing feature fusion set of quantitative parameter arrays introduced here are not mentioned.

[0111] Skin images depicting hematoxylin and eosin stained tissue sections are diagnosed by a professional pathologist as solar keratosis histologically graded KIN (intraepithelial neoplasia of keratinocytes) I, II, or III, or moderately differentiated squamous cell carcinoma (SCC G2)

[11] . Solar keratosis is the most common premalignant lesion in humans, while SCC accounts for approximately 20% of non-melanoma skin cancers and is the second most common type after basal cell carcinoma. Other images depict tissue sections diagnosed by a professional pathologist as Morbus Bowen, an early form of squamous cell carcinoma that is limited to the outermost skin (epidermis).

[0112] Colon images depicting biopsy tissue sections are diagnosed by a professional pathologist as colon adenoma (benign nature of a colon tumor that is considered a precancerous lesion) or colon carcinoma (malignant colon tumor). Note that if colon carcinoma is diagnosed at a localized stage, the 5-year survival rate is 91%, if the cancer has spread to surrounding tissues / organs / regional lymph nodes, the 5-year survival rate is 72%, and if the cancer has spread to distant parts of the body, the 5-year survival rate is 13%. These numbers illustrate the need for early diagnosis and localization, which is usually accomplished by biopsy for histopathological examination using some of the state-of-the-art parameters described above.

[0113] Some specific examples of classification / diagnosis supported by AI / ML learning using the biomarker array are as follows.

[0114] Specific Example I (skin cancer):

[0115] Keratosis actinica (KIN I-III), also known as solar keratosis (solar damaged skin), is a precancerous area of thick, scaly, or leathery skin that is considered a precursor to squamous cell carcinoma (SCC) and can progress to SCC over time. SCC is the second most common form of skin cancer and usually develops from areas of the body that have been chronically exposed to sunlight. Here, we analyzed using scans of pathology slide images. Using a random forest classification model and images diagnosed by a professional pathologist, a universal biomarker array to differentiate KIN I-III from SCC resulted in a correct diagnostic accuracy of 89.64%. See Figure 1 - Random forest classification using a universal biomarker array for diagnosing KIN from SCC, where the upper left and lower parts give multiple parameters of model performance, and the right part gives SHAP (SHapley Additive exPlanation) values, which give the impact of each array parameter on the prediction. The test images were not included in the training data set. Note that these results were achieved with a relatively small number of images (n<100) in each class, which indicates that a larger number of curated and classified images would improve the accuracy.

[0116] Specific Implementation II (colon cancer):

[0117] A colon adenoma is a type of polyp. Up to 10% of colon adenomas can turn into cancer, and thus are precursor lesions for colorectal adenocarcinoma (colon cancer). Here, we analyzed using scans of pathology slide images.

[0118] Using a LightGBM classification model and images diagnosed by a professional pathologist, a universal biomarker array to differentiate colon adenomas from colon cancer produced a correct diagnostic accuracy of 96.53%. See Figure 2 - Predictive classification using a universal biomarker array for diagnosing colon adenomas from cancer, where the upper left and lower parts give multiple parameters of model performance, and the right part gives SHAP (SHapley Additive exPlanations) values, which give the impact of each array parameter on the prediction. The test images were not included in the training data set. Note that these results were achieved with a relatively small number of images (n<100) in each class, which indicates that a larger number of curated and classified images would improve the accuracy.

[0119] DETAILED DESCRIPTION III (Cartilage Degradation):

[0120] Cartilage degradation in osteoarthritis can be identified by spatial organization of superficial chondrocytes (SCSO), a coined term and surrogate marker for loss of tissue functionality (e.g. loss of nanoscale stiffness). Thus, identifying specific stages of SCSO and answering whether a given SCSO is typical of healthy joint cartilage or not allows for classification of health vs. early disease (cartilage degradation). Here, we generated images of tissue biopsies / explants using fluorescence microscopy for analysis.

[0121] Using a random forest classification model and diagnostic images, a universal biomarker array for distinguishing SCSO[healthy] vs. SCSO[diseased] yielded a correct diagnostic accuracy of 91.15% (see Figure 1 ). Test images were not included in the training data set. Note that these results were achieved with a relatively small number of images (n<100) in each class, which indicates that a larger number of curated and classified images would improve accuracy.

[0122] Below is a description of some potential future applications envisioned by the inventors:

[0123] • Development of a clinically applicable, commercially viable tool using the universal biomarker array and associated methods that supports medical decisions in diagnosis and treatment during medical procedures, in particular early disease that has so far been undetectable or difficult to detect. The tool aims to provide real-time and user-friendly feedback to enable surgeons, medical specialists and other experts during medical procedures to diagnose early disease onset and locate early disease processes in tissue earlier than is currently clinically possible, e.g. in combination with microscopic endoscopy or other suitable imaging modalities during medical procedures,

[0124] • Another exemplary application of the tool is to locate early / late disease processes in tissue during medical procedures for assisting in guiding biopsies, assessing / ensuring tumor resection, tissue debridement and implantation of biomaterials / ATMPs / RMATs (FDA, EMA) for regenerative and other purposes, e.g. as described above in combination with microscopic endoscopy or other suitable imaging modalities during medical procedures,

[0125] • Using the universal biomarker array and related methods to develop a fully clinically applicable, commercially available tool that supports medical decision making in diagnosis and staging, which is intended for use by pathologists and other experts evaluating medical images of organs / tissues / tissue sections. One example is implementation through an imaging device and a designated software interface that streamlines images to us for subsequent analysis and testing and provides pathologists or other experts with AI supported diagnosis and staging recommendations. Examples include but are not limited to diseases whose diagnosis requires a biopsy, such as the diseases listed in this document,

[0126] • Using the universal biomarker array and related methods in conjunction with lymph node / bone marrow biopsy, peripheral blood smears, bronchoalveolar lavage, fluid aspiration, brain biopsy, and other biopsy types for cell analysis and analysis described herein,

[0127] • Using the universal biomarker array and related methods in in vitro studies, animal studies, and clinical trials to generate databases with quantified organ, tissue, and / or tissue section specific and health and disease state specific information that are used as reference data for statistical and / or AI / ML supported testing of images / data of interest against the database based reference data to allow for the above applications. In this case, the databases are quantitative maps / digital fingerprints of healthy and diseased organ / tissue / tissue section architectures,

[0128] • Using the universal biomarker array and related methods in conjunction with molecular imaging (e.g., via endomicroscopy as well as PET, SPECT, MRI, optical imaging, ultrasound imaging, photoacoustic imaging, CT, and other imaging modalities) and markers of interest (e.g., antibodies, antibody fragments, small molecules, and other sensors or substances, including commercially available and / or clinically applicable markers) to correlate molecular imaging with early diagnosis, staging, monitoring, and disease localization in organs, tissues / tissue sections generated by the methods described herein,

[0129] • Using the universal biomarker array and related methods on molecular imaging signals / data, e.g., by analyzing markers of interest (e.g., antibodies, antibody fragments, small molecules, and other sensors or substances, including commercially available and / or clinically applicable markers) recorded by molecular imaging to generate quantitative data that can be used as AI / ML input / for statistical analysis. Examples of molecular marker targets are metabolic, inflammatory, and autoimmune processes and / or anatomical and / or pathological structures,

[0130] • Using the above database to correlate information obtained on disease diagnosis and localization by the universal biomarker array (see above) with other data measured in animal studies and clinical trials at the molecular, subcellular, cellular, tissue, organ and patient (e.g. patient reported outcomes) level. An example of an application is, but not limited to, bioinformatics analysis and testing of promising / existing / new / under development drugs targeting specific / early disease processes, e.g. testing existing drugs at early stages of disease, or developing new drugs targeting early stages of disease not previously clinically visible,

[0131] • Using the universal biomarker array and related methods in combination with omics analysis, e.g. genomics, transcriptomics, including spatial transcriptomics, proteomics, metabolomics, radiomics and other fields, to correlate omics data with early diagnosis, staging, monitoring and disease localization in organs, tissues / tissue sections generated by the methods described herein, and if applicable, using the universal biomarker array and related methods on these signals / data,

[0132] • Using the universal biomarker array and related methods in combination with other experimental data of interest, e.g. another specific readout of a given in vitro or ex vivo / in vivo experiment, to determine their statistical relationship. Using this relationship to predict / determine the experimental data of interest by using the universal biomarker array parameter or parameter subset as a surrogate marker. An example is the prediction / determination of OARSI scores or degree of surface damage of articular cartilage from the parameter or parameter subset (after statistical relationship is established by appropriate analysis),

[0133] • Using the universal biomarker array and related methods on in vitro cell, organ, tissue cultures as quantitative markers and predictors of cell, organ, tissue culture performance (e.g. cell phenotype, differentiation and other functions),

[0134] • Using the universal biomarker array and related methods on all mammalian species / tissues / cells, e.g. human volunteers, human patients and animals such as dogs, horses and camels, for the above purposes.

[0135] • Examples of diseases for the above listed applications include, but are not limited to

[0136] • Cancer and precancerous lesions, e.g.

[0137] o skin,

[0138] o gastrointestinal,

[0139] o lung,

[0140] o kidney,

[0141] o breast,

[0142] • liver,

[0143] • esophagus,

[0144] • brain,

[0145] • prostate,

[0146] • pancreas,

[0147] • glioblastoma,

[0148] • metastasis,

[0149] • orthopedic conditions / diseases, including but not limited to

[0150] • injury to cartilage and other joint structures,

[0151] • degeneration of cartilage and other joint structures,

[0152] • genetic conditions involving cartilage,

[0153] • osteochondritis dissecans,

[0154] • inflammatory arthropathies (rheumatoid arthritis, juvenile idiopathic arthritis, gout, systemic lupus erythematosus, and seronegative spondylarthritides),

[0155] • osteoarthritis,

[0156] • post-traumatic osteoarthritis,

[0157] • dysplasia of the cartilage

[0158] • vascular diseases, for example,

[0159] • atherosclerosis,

[0160] • stenosis,

[0161] • hypertension,

[0162] • stroke

[0163] • heart diseases, for example,

[0164] • coronary artery disease / atherosclerosis,

[0165] • cardiomyopathy,

[0166] • myocarditis,

[0167] • heart transplant rejection,

[0168] • gastrointestinal diseases, for example.

[0169] • polyps,

[0170] • gastritis / brain ventricle ulcers,

[0171] o Unclear mass / lesion of the gastrointestinal tract,

[0172] o Cancer,

[0173] • Kidney diseases, e.g.

[0174] o Chronic kidney disease,

[0175] o Glomerulonephritis,

[0176] o Renal fibrosis,

[0177] • Liver diseases, e.g.

[0178] o Liver cirrhosis,

[0179] • Inflammatory / autoinflammatory / autoimmune diseases,

[0180] • Eye diseases, e.g.

[0181] o Glaucoma,

[0182] o Macular degeneration,

[0183] • Neurodegenerative diseases, e.g.

[0184] o Parkinson’s disease,

[0185] o Alzheimer’s disease,

[0186] o Huntington’s disease,

[0187] o Amyotrophic lateral sclerosis (ALS),

[0188] o Motor neuron disease,

[0189] • Chronic inflammatory brain diseases, e.g.

[0190] o Multiple sclerosis,

[0191] o Acute demyelination,

[0192] • Vascular pathologies diseases,

[0193] o e.g. cerebral amyloid angiopathy (CAA)

[0194] • Peripheral nerve injury

[0195] • Rare diseases,

[0196] • Metabolic disorders / diseases, e.g.

[0197] o Diabetes

[0198] The following is information on prior publications (reference numbers in brackets refer to the reference citations listed below). It should be noted that all of these studies involved only cartilage cells / cartilage / osteoarthritis (OA). In studies from 2008 to 2014, we used the term “spatial organization” and referred only to cartilage cells / cartilage / osteoarthritis (OA). These analyses used very limited quantitative data, e.g., with 1-2 markers as part of the universal biomarker array described herein (Clark Evans exponent, nearest neighbor distance). Since 2016, we have used the term “superficial zone cartilage cell spatial organization (SCSO).” A small number of quantitative parameters were used in some later studies of articular cartilage. Some of the terms used, related measurements, and key statements related to disease are highlighted in bold to illustrate the very limited use of a small subset of the universal biomarker array and limited context (only cartilage, cartilage cells, degeneration, osteoarthritis). It should be noted that certain cell morphometric parameters have been used to describe quantitative effects of stimulation but not in conjunction with diagnosis, etc.

[0199] • Different patterns of levels in the superficial zone cartilage cell spatial organization of human joints [1]:

[0200] A better understanding of the unique cellular and functional properties of the superficial zone of articular cartilage can help current strategies in tissue engineering that attempt a stratified design for cartilage injury repair to avoid or delay the onset of osteoarthritis. However, data on the cellular organization of the non-degenerate superficial zone of articular cartilage are not available for most human joints. This study analyzed the arrangement of non-degenerate human articular (shoulder, elbow, knee, and ankle) cartilage cells using a top-down superficial fluorescence microscopy. The resulting horizontal cartilage cell arrangements were tested for randomness, homogeneity, or significant grouping by point pattern analysis and related to the joint type in which they occurred. This study showed that human superficial cartilage cells occur in four different patterns, namely strings, clusters, pairs, or single cartilage cells. These patterns represent significant groupings of horizontal alignment (p < 0.0001). Each joint surface is dominated by only one of these four patterns (p < 0.001). Specific patterns are related to specific diarthrodial joint types (p < 0.001). Further studies are needed to determine whether these organizational patterns are a result of the surrounding environment or are related to functional purposes.

[0201] • Proliferative remodeling of human superficial cartilage cell spatial organization away from focal early osteoarthritis [2]:

[0202] Objective: Human superficial zone chondrocytes exhibit a unique spatial organization, which they usually aggregate around osteoarthritic (OA) fissures. The aim of this study was to determine whether remodeling or disruption of the spatial chondrocyte organization in OA patients can occur at a distance from the focal (early) lesion. Methods: Intact cartilage (condyle, patellofemoral groove, and proximal tibia) samples from grade 2 joints, distal to the OA focal lesion, were compared with position-matched non-degenerative (grade 0-1) cartilage samples. Chondrocyte nuclei were stained with propidium iodide, examined by fluorescence microscopy, and findings were recorded in top-down views. Point pattern analysis (Clark and Evans aggregation index) was used to test for randomness or significant grouping of chondrocyte arrangements and correlation with OA grade and superficial cell density. Results: In grade 2 cartilage samples, superficial zone chondrocytes were organized in horizontal patterns, such as strings, clusters, pairs, and singles, comparable to those in non-degenerative cartilage. In intact cartilage samples from grade 2 joints, the spatial organization included a novel pattern, consisting of chondrocytes arranged in 2 parallel lines, forming double strings. These double strings were significantly associated with an increase in the number of chondrocytes per group and a corresponding increase in superficial zone cell density. They were observed in all grade 2 condyles and some grade 2 tibiae but never in grade 0-1 cartilage. Conclusion: This study first identified a unique spatial reorganization of human superficial zone chondrocytes to a distant early OA lesion, indicating that proliferation occurs at a distance from the focal early OA lesion. This spatial reorganization can serve to recruit metabolically active units as an attempt to repair the focal lesion.

[0203] • Preclinical onset of osteoarthritis: angular spatial organization allows early diagnosis[3]:

[0204] Objective: Superficial articular chondrocytes exhibit different spatial remodeling processes in distal osteoarthritis (OA). Such processes can be used to diagnose early events before overt OA leads to tissue destruction and clinical symptoms. Using a novel spatial quantification method by calculating the angles between chondrocytes and their surrounding neighbors, we compared the maturation and degenerative changes in cell organization in rat and human cartilage specimens. Methods: Superficial chondrocyte nuclei obtained from intact rat cartilage and human knee cartilage as well as cartilage with focal and severe OA were digitally recorded in a top-down fashion. Their Cartesian coordinates were used to determine the nearest neighbors of each chondrocyte and the angles between these two cells and a reference. These angles, cell structure, nearest neighbor distance, and aggregation were analyzed as a function of location and OA severity. Results: Adjacent rat chondrocytes exhibit a complex angular pattern that maintains 4 dominant angles during maturation and during the onset and progression of OA. Human chondrocytes exhibit 1 dominant angle in intact cartilage and thus a significantly different angular organization. At the onset of early OA, human chondrocytes located within intact cartilage show an increased incidence of 4 angles; the resulting angular pattern is indistinguishable from that observed in rats. The remodeling of angles is associated with changes in the location of cell structure and aggregation and OA severity. Conclusion: This study first identifies the existence of angular features of spatial chondrocyte organization and species-specific remodeling processes associated with the onset of OA. The emergence of different angular and spatial patterns between adjacent chondrocytes can identify the onset of distal OA before tissue damage visible under a microscope and possibly before clinical onset. With further development, this novel concept can be applied to the diagnosis and follow-up of patients predisposed to OA.

[0205] • Chondrocyte patterns are modeled by an elliptical cluster process[4]:

[0206] Superficial chondrocytes of human joints (CHs) are spatially organized in different levels of patterns. Among other factors, the type of spatial CH organization within a given joint surface depends on whether the cartilage is from an intact joint or whether the joint is affected by osteoarthritis (OA). Moreover, specific changes in the type of spatial organization are associated with specific states of OA. Based on the quantitative structural characterization of CH patterns, this association can prove relevant to early disease identification. Therefore, we propose a point process model that describes the different morphologies of CH patterns within intact human cartilage joint surfaces. This reference model for intact CH organization can be considered the first step toward a model-based statistical diagnostic tool. The model parameters are fitted to fluorescence microscopy data by using a novel statistical method that leverages clustering and principal component analysis tools. In this way, the complex morphologies of surface CH patterns are represented by a relatively small number of model parameters. We validate the point process model by comparing the biologically relevant structural features between the fitted model and data obtained from human joint surface micrographs using spatial statistical techniques.

[0207] • Spatial organization of articular surface chondrocytes: a review of its potential role in tissue function, disease, and early preclinical diagnosis of osteoarthritis [5]:

[0208] Chondrocytes show deep interrelated changes in their many properties within articular cartilage, comparable to deep interrelated changes in properties of the surrounding extracellular matrix. However, little is known about the spatial organization of chondrocytes in the whole tissue. Recent studies have shown that human chondrocytes show different spatial organization patterns within the articular surface and that each articular surface is dominated by one of four basic spatial patterns in a typical manner. The resulting complex spatial organization is related to specific diarthrodial joint types, suggesting that chondrocyte organization within the articular surface is related to the occurring biomechanical forces. In response to focal osteoarthritis (OA), superficial chondrocytes undergo a disruption of their spatial organization within the OA lesion, but they also undergo a defined remodeling process in the remaining intact cartilage surface away from the OA lesion. One of the biological insights that can be drawn from this spatial remodeling process is that chondrocytes are able to react to distant focal OA in a universal and coordinated manner. The spatial characteristics of this process are quite different from the typical cell aggregation of OA lesions, suggesting differences in underlying mechanisms. Here, we summarize the available information on chondrocyte spatial organization and its potential role in cartilage function. Spatial organization can be used to diagnose early onset of OA before obvious OA causes tissue destruction and clinical symptoms. With further development, this concept can be clinically applicable for preclinical diagnosis of OA.

[0209] • (Meeting abstract) Spatial organization and pericellular matrix of human fetal chondrocytes are not innate but acquired features [6]:

[0210] Introduction. In adult articular cartilage, the pericellular matrix mediates biomechanical, biophysical and biomechanical interactions between chondrocytes and extracellular matrix. PCM is also related to the spatial organization of human superficial chondrocytes, which are located in four different patterns: strings, clusters, pairs or single cells. However, little is known about PCM and spatial organization during fetal development. In this study, we addressed the question of whether fetal chondrocytes exhibit a spatial organization comparable to that of adult chondrocytes and whether PCM is present in early stages of fetal cartilage development. Articular cartilage sections (100 µm thick) were prepared from human fetal knee condyles (gestational age 7-10 weeks) and from macroscopically intact condylar regions (knee replacement surgery). Samples were characterized by immunofluorescence microscopy and multiphoton-induced autofluorescence imaging combined with quantitative SHG signal profiling, a technique that allows quantification of fibrillar collagen content. Spatial organization was analyzed by point pattern analysis as previously described. For these analyses, the Cartesian coordinates of each nucleus were determined by converting the immunofluorescence images into grayscale images and finding the local grayscale maxima with ImageJ (NIH). In adult superficial cartilage, PCM surrounds individual or groups of chondrocytes and defines their spatial organization in cell string groups. PCM is characterized by strong collagen VI staining signal and high collagen intensity as measured by SHG (156.7 ± 12.4). In fetal cartilage, the condyle is characterized by high cell density, no recognizable spatial organization and low amounts of extracellular matrix. No collagen VI was detected in the matrix surrounding fetal chondrocytes. Moreover, SHG imaging revealed significantly weaker fibrillar collagen intensity (4.7 ± 0.8; p < 0.001). Cell density per volume significantly decreased during maturation from fetal to adult stage (p < 0.001). The cell distance of each fetal chondrocyte to its nearest neighbor was 15.70 ± 0.12 µm, significantly larger in adult cartilage (35.86 ± 1.37 µm; p < 0.001). The level of spatial clustering as measured by integration of the pair correlation function was significantly higher in adult cartilage (p < 0.001). Overall, no cell grouping comparable to that of adult chondrocytes was present in fetal cartilage. These parameters collectively indicate that the spatial organization typical of adult condylar cartilage is not present in human fetal condylar cartilage. Discussion / conclusion In human fetal articular cartilage, no spatial organization typical of adult chondrocytes is present. Instead, chondrocytes are densely packed and close together. This study first determined that the collagenous composition of PCM typical of adult PCM is not present in fetal cartilage. In summary, PCM as well as spatial organization of superficial human chondrocytes develops with cartilage maturation and is thus not an innate but an acquired feature. In adult articular cartilage, the pericellular matrix (PCM) mediates chondrocyte-matrix-interactions and is related to spatial cell organization.Immunofluorescence microscopy, multiphoton-induced autofluorescence and second harmonic generation (SHG) imaging and point pattern analysis revealed the absence of PCM and spatial organization in fetal chondrocytes.

[0211] • Loss of spatial organization and pericellular matrix destruction in early osteoarthritis in vivo and in a novel ex vivo method [7]:

[0212] Objective: Current articular cartilage (AC) repair procedures cannot restore the original form and function of the tissue because its structural blueprint changes throughout life and the respective biological understanding is not fully available. We asked whether two unique elements of human cartilage structure, the pericellular matrix (PCM) around chondrocytes and the spatial organization of superficial zone chondrocytes (SCSO) under the articular surface (AS), are innate, stable or dynamic throughout life. We hypothesized that inducing chondrocyte proliferation in vitro would impair the tissue and PCM and induce an osteoarthritis (OA)-like structural phenotype in human cartilage. Methods: We recorded propidium iodine-stained fetal and adult cartilage explants, arranged the tissue stages into a sequence and created a life-long summary SCSO model. To replicate the OA-related dynamics our model revealed and to test our hypothesis, we specifically transduced early OA explants with hFGF-2 to induce proliferation. The PCM was examined using immunofluorescence and autofluorescence, multiphoton second harmonic generation (SHG) and scanning electron microscopy (SEM). Results: Spatial organization evolved from fetal homogeneity to peak in adult string-like arrangement but was completely lost in OA. Tissue loss included PCM perforation (local microfibrillar collagen strength reduction) and destruction

local type VI collagen (CollVI) signal weakness or absence

[0213] • Proof of concept for early osteoarthritis pathology detection by clinically applicable microscopic endoscopy and quantitative artificial intelligence-supported optical biopsy [8]:

[0214] Objective: Osteoarthritis (OA) is a leading cause of disability worldwide, and its clinical trials cannot identify early, potentially reversible disease by clinical techniques. Therefore, candidate drugs to improve the disease cannot be tested in the early stages of the disease. To overcome this obstacle, we asked whether early OA pathology can be detected with current clinical techniques. Methods: We identified two sensitive markers of early OA, the relationship between human articular cartilage (AC) surface stiffness measured by atomic force microscopy (AFM) and spatial organization of superficial zone chondrocytes (SCSO) in position-matched AC, and asked whether a significant loss of surface stiffness can be detected at the SCSO stage of early OA. Then, we tested whether current clinical techniques can visualize and accurately diagnose SCSO using an approved probe-based confocal laser endomicroscopy and random forest (RF) model. Results: We demonstrated the correlation between AC surface stiffness and SCSO (r(rm) = -0.91; 95% CI: -0.97, -0.73) and the widespread loss of surface stiffness, particularly in AC with SCSO typical of early OA (95% CI: single SCSO: 269-173 kPa, double SCSO: 77-46 kPa). This established SCSO as a visualizable, functionally relevant surrogate marker of early OA AC surface pathology. Moreover, SCSO-based stiffness discrimination worked well in each patient's AC. Then, we demonstrated the feasibility of observing SCSO by clinical laser endomicroscopy and, more importantly, accurate SCSO diagnosis using radiofrequency. Conclusion: We present a proof of concept for early OA pathology detection using existing clinical techniques, introducing a future-oriented, artificial intelligence-supported, non-destructive quantitative optical biopsy for early disease detection. With SCSO recognition, this approach allows testing of the correlation between local tissue architecture and other experimental and clinical readouts, but requires clinical validation and larger sample sizes to define diagnostic thresholds.

[0215] • Mixed fluorescence-AFM explores joint surface degeneration in early osteoarthritis across length scales[9]:

[0216] Atomic force microscopy (AFM) has become a powerful tool for characterizing nanoscale materials. However, its application in hierarchical biological tissues such as cartilage remains limited. One reason is that such samples are often millimeter-sized, while AFM provides more local information. Here we propose a combination of AFM and fluorescence microscopy, where features on a millimeter-sized tissue sample are selected by fluorescence microscopy on a micrometer scale, and then mapped by AFM to nanometer precision under native conditions. This helped us to show that local changes in fluorescence-stained cells, a marker of early osteoarthritis, are associated with a significant local decrease in elastic modulus, local thinning of collagen fibers, and roughening of the articular surface. This approach is not only relevant to cartilage but also to the characterization of natural biological tissues from the macro- to nanoscale in general. Significance statement: Different length scales must be investigated to understand the function and dysfunction of hierarchical biological materials or tissues. Here we combine highly stable AFM with fluorescence microscopy and precise motorized motion to correlate micrometer and nanometer properties of articular cartilage on millimeter-sized samples under native conditions. This is necessary to reveal the relationship between the microscale organization of chondrocytes, the micrometer-scale variations in articular cartilage properties, and the nanoscale organization of collagen, including the D-bands. We anticipate that such studies will pave the way for guided design of hierarchical biological materials.

[0217] • In early osteoarthritis, the articular cartilage surface is compromised by loss of thick collagen fibers and formation of type I collagen

[10] :

[0218] Osteoarthritis (OA) is a joint disease affecting millions of patients worldwide. During the onset and progression of OA, articular cartilage is destroyed, but the underlying complex mechanisms remain unclear. Here, we reveal changes in collagen fiber thickness and its composition at the onset of osteoarthritis. For articular cartilage explants from knee joints of OA patients, we found that fibrocartilage-like tissue rich in type I collagen forms in macroscopically intact cartilage away from OA lesions. Importantly, the number of thick fibers (>100 nm) is already reduced in early-stage disease and completely absent in late-stage OA. We obtained these results by combining high-resolution atomic force microscopy imaging under near-native conditions, immunofluorescence, scanning electron microscopy, and fluorescence-based spatial organization classification of superficial zone chondrocytes. Taken together, our data suggest that the loss of tissue function in early-stage OA cartilage is caused by a reduction in thick type II collagen fibers, possibly due to the formation of fibrocartilage rich in type I collagen, followed by the development of focal defects in late-stage OA. We anticipate that this comprehensive characterization will be highly beneficial for gaining insight into other natural biological tissues and for developing sustainable biomaterials. Significance Statement: In early-stage osteoarthritis (OA), cartilage appears macroscopically intact. However, this study shows that in early-stage OA, the collagen network has already changed due to a thinning of collagen fibers and the formation of fibrocartilage-like tissue. These two nanoscale defects have occurred in macroscopically intact regions of human knee joints and can be related to processes that lead to weakening of the extracellular matrix. This study enhances the understanding of the earliest progressive cartilage degeneration without external damage. The results suggest that measuring average collagen fiber thickness as a new target for detecting early-stage OA and modulating type I collagen synthesis as a new approach for OA treatment.

[0219] The following references were used in the above discussion.

[0220] [1] Rolauffs, B., et al., Distinct horizontal patterns in the spatial organization of superficial zone chondrocytes of human joints. Journal of Structural Biology, 2008. 162(2): p. 335-344.

[0221] [2] Rolauffs, B. et al. Proliferative remodeling of the spatial organization of human superficial chondrocytes distant from focal early osteoarthritis. Arthritis Rheum, 2010. 62(2): p. 489-498.

[0222] [3] Rolauffs, B. et al. Onset of preclinical osteoarthritis: the angular spatial organization permits early diagnosis. Arthritis Rheum, 2011. 63(6): p. 1637-1647.

[0223] [4] Meinhardt, M. et al. Modeling chondrocyte patterns by elliptical cluster processes. 2012. 177(2): p. 447-458.

[0224] [5] Aicher, W.K. and B. Rolauffs, The spatial organisation of joint surface chondrocytes: review of its potential roles in tissue functioning, disease and early, preclinical diagnosis of osteoarthritis. Ann Rheum Dis, 2014. 73(4): p. 645-653.

[0225] [6] Felka, T. et al. THE SPATIAL ORGANISATION AND THE PERICELLULAR MATRIX OF HUMAN FOETAL CHONDROCYTES ARE NOT INBORN BUT INSTEAD ACQUIRED CHARACTERISTICS. Bone & Joint Journal Orthopaedic Proceedings Supplement, 2014. 96-B (Suppl 11): p. 323.

[0226] [7] Felka, T. et al. Loss of spatial organization and destruction of the pericellular matrix in early osteoarthritis in vivo and in a novel in vitro methodology. Osteoarthritis Cartilage, 2016. 24(7): p. 1200-1209.

[0227] [8] Tschaikowsky, M. et al. Proof-of-concept for the detection of early osteoarthritis pathology by clinically applicable endo-microscopy and quantitative AI-supported optical biopsy. Osteoarthritis Cartilage, 2021. 29(2): p. 269-279.

[0228] [9] Tschaikowsky, M. et al. Hybrid fluorescence-AFM explores articular surface degeneration in early osteoarthritis across length scales. Acta Biomater, 2021. 126: p. 315-325.

[0229]

[10] Tschaikowsky, M. et al. The articular cartilage surface is impaired by a loss of thick collagen fibers and formation of type I collagen in early osteoarthritis. Acta Biomaterialia, 2022. 146: p. 274-283.

[0230]

[11] Yantsos, V. et al. Incipient intraepidermal cutaneous squamous cell carcinoma: a proposal for reclassifying and grading solar (actinic) keratoses. Semin Cutan Med Surg 1999; 18(1): 3-14. (English). DOI: 10.1016 / s1085-5629(99)80003-0.

[0231] Various embodiments of the present application can be implemented, at least in part, in any conventional computer programming language. For example, some embodiments can be implemented in a procedural programming language (e.g., "C"), or in an object oriented programming language (e.g., "C++"). Other embodiments of the present application can be implemented as preconfigured standalone hardware elements and / or preprogrammed hardware elements (e.g., application specific integrated circuits, FPGAs, and digital signal processors) or other related components.

[0232] In alternative embodiments, the disclosed apparatus and methods (e.g., as in any of the flowcharts or logic flow diagrams above) can be implemented as a computer program product for use with a computer system. Such implementation can include a series of computer instructions fixed either on a tangible, non-transitory medium, such as a computer readable medium (e.g., a diskette, CD-ROM, ROM, or hard disk) or transitory medium, such as a propagating signal, for example. The series of computer instructions can embody all or part of the functionality previously described herein with regard to the system.

[0233] Those skilled in the art should appreciate that such computer instructions can be written in a number of programming languages for use with many computer architectures or operating systems. Furthermore, such instructions can be stored in any memory device, such as a tangible, non-transitory semiconductor, magnetic, optical or other memory device, and can be transmitted using any communications technology, such as optical, infrared, RF / microwave or other transmission technologies, for example, over any appropriate medium (e.g., wired (e.g., electrical wire, coaxial cable, fiber optic cable, etc.) or wireless (e.g., through air or space)).

[0234] In other approaches, such a computer program product can be distributed as a removable medium with accompanying printed or electronic documentation (e.g., shrink wrapped software), preloaded with a computer system (e.g., on a system ROM or fixed disk), or distributed from a server or electronic bulletin board over the network (e.g., the Internet or World Wide Web). Or, some embodiments can be implemented as a service (e.g., software as a service, or "SAAS") or cloud computing scenario. Of course, some embodiments of the application can be implemented as a combination of both software and hardware. Still other embodiments are implemented as entirely hardware, or entirely software.

[0235] Computer program logic implementing all or part of the previous described functionality can be executed by a single processor (e.g., concurrently), or can be executed by multiple processors at the same or different times, and can run under a single operating system process / thread or under different operating system processes / threads. Thus, the term "computer process" generally refers to the execution of a set of computer program instructions by one or more processors, regardless of whether different computer processes are executing on the same or different processors, and regardless of whether different computer processes are running under the same or different operating system processes / threads. A software system can be implemented using a variety of architectures; for example, object-oriented architectures, relational architectures, etc.

[0236] Importantly, it should be noted that embodiments of the present application can employ conventional components, such as conventional computers (e.g., commercially available PCs, mainframes, microprocessors), conventional programmable logic devices (e.g., commercially available FPGAs or PLDs), or conventional hardware components (e.g., commercially available ASICs or discrete hardware components), which when programmed or configured to perform the unconventional methods described herein, result in unconventional devices or systems. Thus, there is nothing conventional about the inventions described herein, as even when embodiments are implemented using conventional components, the resulting devices and systems are necessarily unconventional, as the conventional components do not inherently perform the unconventional functions described without being specially programmed or configured.

[0237] The activities described and claimed herein provide technical solutions to problems that have arisen in the technical field. These solutions are not well understood, conventional, or routine as a whole and provide practical applications that improve computer and computer routing systems in any case.

[0238] While various inventive embodiments have been described and illustrated herein, those of ordinary skill in the art will readily envision a variety of other means and / or structures for performing the functions and / or obtaining the results and / or one or more of the advantages described herein, and each of such variations and / or modifications is deemed to be within the scope of the inventive embodiments described herein. More generally, those skilled in the art will readily appreciate that all parameters, dimensions, materials, and configurations described herein are meant to be exemplary and that the actual parameters, dimensions, materials, and / or configurations will depend upon the specific application or applications for which a inventive embodiment is used. Those skilled in the art will recognize, or be able to ascertain using no more than routine experimentation, many equivalents to the specific inventive embodiments described herein. It is, therefore, to be understood that the foregoing embodiments are presented by way of example only and that inventive embodiments can be practiced within a wide and equivalent range of conditions, materials, and / or configurations. Inventive embodiments herein relate to each individual feature, system, article, material, kit, and / or method described herein. In addition, any combination of two or more such features, systems, articles, materials, kits, and / or methods, if such features, systems, articles, materials, kits, and / or methods are not mutually inconsistent, is included within the inventive scope of the present disclosure.

[0239] Various inventive concepts can be embodied in one or more methods, of which an example has been provided. The acts performed as part of the method can be ordered in any suitable way. Accordingly, an inventive embodiment can be constructed in which acts are performed in an order different than illustrated, which can include performing some acts simultaneously, even though shown as sequential acts in illustrative embodiments.

[0240] All definitions, as defined and used herein, should be understood to control over dictionary definitions, definitions in documents incorporated by reference, and / or ordinary meanings of the defined terms.

[0241] The indefinite articles "a" and "an," as used herein in the specification and in the claims, unless clearly indicated to the contrary, should be understood to mean "at least one."

[0242] As used herein in the specification and claims, the phrase “and / or” should be understood to mean “any one or both” of the elements so combined, i.e., elements that exist together in some cases and separately in others. Multiple elements listed with “and / or” should be interpreted in the same way, i.e., “one or more” of the elements so combined. In addition to the elements specifically identified by the “and / or” clause, other elements may optionally be present, whether related to or unrelated to those specifically identified. Thus, as a non-limiting example, when used in conjunction with open-ended language such as “comprising,” a reference to “A and / or B” may in one embodiment refer only to A (optionally including elements other than B); in another embodiment, only to B (optionally including elements other than A); in yet another embodiment, to both A and B (optionally including other elements); and so on.

[0243] As used herein in the specification and claims, “or” should be understood to have the same meaning as “and / or” as defined above. For example, when separating items in a list, “or” or “and / or” should be interpreted as inclusive, i.e., including multiple elements or at least one of the items in the list, but also including more than one, and optionally, additional unlisted items. Only terms that explicitly indicate the opposite, such as “only one” or “exactly one”, or when used in a claim, “consisting of” will refer to including exactly one of a plurality of elements. In general, the term “or” as used herein should only be interpreted as indicating an exclusive alternative (i.e., “one or the other but not both”) when preceded by an exclusive term, such as “any one,” “one of,” “only one of,” or “exactly one of.” When used in a claim, “consisting substantially of” should have its ordinary meaning as used in the field of patent law.

[0244] As used herein in the specification and claims, the phrase “at least one,” in reference to a list of one or more elements, should be understood to mean at least one element selected from among the individual elements in the list of elements, and is not intended to be construed to mean that the listed elements are required to some combination total number of elements in the list. It is not intended that the phrase “at least one” be construed in any way that would impart a meaning that is not consistent with the plain meaning of the phrase as used in the specification and claims.

[0245] As used herein in the specification and claims, all transitional phrases such as “comprising,” “including,” “carrying,” “having,” “containing,” “involving,” “holding,” “composed of,” and the like are to be understood to be open-ended, i.e., to mean including but not limited to. Only the transitional phrases “consisting of’ and “consisting essentially of’ shall be closed or semi-closed transitional phrases, as set forth in the United States Patent Office Manual of Patent Examining Procedures, Section 2111.03.

[0246] Various implementations of the application can be characterized in potential claims listed in the paragraphs that follow this paragraph (and prior to the actual claims provided at the end of this application). These potential claims form part of the written description of this application. As such, the subject matter of the following potential claims can be presented as actual claims in a subsequent litigation involving this application or any application claiming priority from this application. The inclusion of such potential claims should not be interpreted as meaning that the actual claims do not cover the subject matter of the potential claims. Therefore, a decision not to present these potential claims in a later litigation should not be interpreted as a donation of subject matter to the public. These potential claims are also not intended to limit the various claims being pursued.

[0247] But without limitation, potential subject matter that can be claimed (prefaced with the letter “P” to avoid confusion with actual claims presented below) includes:

[0248] P1. A method of disease diagnosis, the method comprising: receiving an image of an organ / tissue / tissue section; performing automated image analysis to identify organ / tissue / tissue section / cell in the image from image background, construct a region of interest (ROI) in non-background regions, and segmentation based on ROI to (i) segment cells for analysis of segmented cells and (ii) analyze inverse segmentation for analysis of space between cells (stroma); compute a series of mathematical parameters at multiple levels to aggregate for use as a universal biomarker array; store curated (e.g., organ / tissue / tissue section / cell specific and disease state specific) diagnostic / classification / annotation universal biomarker array data in a database to produce database-based reference data; and perform statistical and / or AI / ML supported testing against the database-based reference data using images / data of interest for early diagnosis and localization of disease processes.

[0249] P2. A system comprising at least one processor and at least one memory including instructions that, when executed by the at least one processor, cause the system to perform processes comprising: receiving an image of an organ / tissue / tissue section; performing automated image analysis to identify organ / tissue / tissue section / cell in the image from image background, construct a region of interest (ROI) in non-background regions, and segmentation based on ROI to (i) segment cells for analysis of segmented cells and (ii) analyze inverse segmentation for analysis of space between cells (stroma); compute a series of mathematical parameters at multiple levels to aggregate for use as a universal biomarker array; store curated (e.g., organ / tissue / tissue section / cell specific and disease state specific) diagnostic / classification / annotation universal biomarker array data in a database to produce database-based reference data; and perform statistical and / or AI / ML supported testing against the database-based reference data using images / data of interest for early diagnosis and localization of disease processes.

[0250] While the above discussion discloses various exemplary embodiments of the application, it should be apparent that those skilled in the art can make various modifications that will achieve some of the advantages of the application without departing from the true scope of the application. Any reference to "the invention" is intended to mean an exemplary embodiment of the application and not all embodiments of the application unless the context is clear. The described embodiments are to be considered in all respects only as illustrative and not restrictive.

Claims

1. A method for disease diagnosis, the method comprising: receiving a digital tissue image showing cellular level tissue detail; performing automated image analysis on the received digital tissue image to identify cells of interest for analysis; computing a series of mathematical parameters at multiple levels to collectively serve as a universal biomarker array, the universal biomarker array including a pattern array, a distance array, a morphology array, a spatial entropy array, a box array in which absolute parameter values are converted to relative information by assigning the absolute parameter values to specific box locations, and a quartile array in which absolute parameter values are converted to relative information by assigning the absolute parameter values to specific quartiles; and analyzing the universal biomarker array against database-based reference data for early diagnosis and localization of disease processes.

2. The method of claim 1, wherein, The digital tissue image includes an endomicroscopy digital tissue image.

3. The method of claim 1, wherein, Performing automated image processing includes identifying cells of interest against an image background, constructing at least one region of interest (ROI) in non-background areas, and performing ROI-based segmentation to (i) segment cells for analysis of the segmented cells and (ii) analyze inverse segmentation for analysis of spaces between the cells.

4. The method of claim 1, wherein, The spatial entropy array parameters include Batty (absolute, relative), Contagion, Karlstrom (absolute, relative), O'Neill (absolute, relative), and Parredw parameters.

5. The method of claim 1, wherein, Computing the box array includes assigning each of a plurality of parameters to relative class-specific box locations and range-specific box locations within an entire range of parameter values, wherein a total number of boxes is computed by multiplying a number of disease states by a number of boxes per state.

6. The method of claim 5, wherein, The conversion of data to relative class-specific box locations and range-specific box locations is performed on parameters of the spatial entropy array, the pattern array, the distance array, and the morphology array.

7. The method of claim 1, wherein, Computing the quartile array includes: assigning each of a plurality of parameter values to a disease state-specific quartile location; and converting each of the plurality of parameter values to a disease range-specific quartile location.

8. The method of claim 7, wherein, The conversion of data to relative class-specific quartile locations and range-specific quartile locations is performed on parameters of the spatial entropy array, the pattern array, the distance array, and the morphology array.

9. The method of claim 1, wherein, Analyzing the universal biomarker array against database-based reference data includes: providing the universal biomarker array to an AI / ML system trained using universal biomarker array data to detect early diagnosis and localization of disease processes.

10. The method of claim 9, wherein, The AI / ML system utilizes random forest regression to detect early diagnosis and localization of disease processes.

11. A system for disease diagnosis, the system comprising: a computer system having at least one computer processor and an associated memory containing computer program instructions that, when executed by the at least one computer processor, perform a computer process, the computer process comprising: receiving a digital tissue image showing cellular level tissue detail; performing automatic image analysis on the received digital tissue image to identify cells of interest for analysis; computing a series of mathematical parameters at multiple levels to aggregate for use as a universal biomarker array, the universal biomarker array including a pattern array, a distance array, a morphology array, a spatial entropy array, a box array in which absolute parameter values are converted to relative information by assigning the absolute parameter values to specific box locations, and a quartile array in which absolute parameter values are converted to relative information by assigning the absolute parameter values to specific quartiles; and analyzing the universal biomarker array relative to database based reference data for early diagnosis and localization of disease processes.

12. The system of claim 11, wherein, the digital tissue image includes an endomicroscopy digital tissue image.

13. The system of claim 11, wherein, performing automatic image processing includes identifying cells of interest relative to image background, constructing at least one region of interest (ROI) in non-background areas, and performing ROI based segmentation to (i) segment cells for analysis of the segmented cells and (ii) analyze inverse segmentation for analysis of spaces between the cells.

14. The system of claim 11, wherein, the spatial entropy array parameters include Batty (absolute, relative), Contagion, Karlstrom (absolute, relative), O'Neill (absolute, relative), and Parredw parameters.

15. The system of claim 11, wherein, computing the box array includes assigning each of a plurality of parameters to relative class specific box locations and range specific box locations within an entire range of parameter values, wherein a total number of boxes is computed by multiplying a number of disease states by a number of boxes per state.

16. The system of claim 15, wherein, the conversion of data to relative class specific box locations and range specific box locations is performed on the parameters of the spatial entropy array, the pattern array, the distance array, and the morphology array.

17. The system of claim 11, wherein, computing the quartile array includes: assigning each of a plurality of parameter values to a disease state specific quartile location; and converting each of the plurality of parameter values to a disease range specific quartile location.

18. The system of claim 17, wherein, the conversion of data to relative class specific quartile locations and range specific quartile locations is performed on the parameters of the spatial entropy array, the pattern array, the distance array, and the morphology array.

19. The system of claim 11, wherein, analyzing the universal biomarker array relative to database based reference data includes: providing the universal biomarker array to an AI / ML system trained using universal biomarker array data to detect early diagnosis and localization of disease processes.

20. The system of claim 19, wherein, the AI / ML system utilizes random forest regression to detect early diagnosis and localization of disease processes.

21. The method of claim 1, wherein, the digital tissue image includes a digital tissue image from a digital pathology slide scanner.

22. The method of claim 1, wherein, the digital tissue image includes a confocal microscope digital tissue image.

23. The method of claim 1, wherein, the digital tissue image includes a fluorescent microscope digital tissue image.

24. The method of claim 1, wherein, the digital tissue image includes a digital tissue image of stained cells.

25. The method of claim 1, wherein, the digital tissue image includes a digital tissue image of cells treated using fluorescent materials.

26. The method of claim 25, wherein, The fluorescent material comprises a fluorescent molecule.

27. The method of claim 25, wherein, The fluorescent material comprises a self-fluorescing drug.

28. The method of claim 27, wherein, The self-fluorescing drug comprises tetracycline, ciprofloxacin or quinine.

29. The method of claim 27, wherein, The self-fluorescing drug comprises doxycycline with divalent cations such as magnesium and isotonic solutions commonly used in surgical procedures.

30. The method of claim 25, wherein, The digital tissue image comprises a digital tissue image of cells treated with a fluorescent material that is excited in the range of 350 to 400 nm (e.g. with UV-A light). The digital tissue image comprises a digital tissue image of cells treated with a fluorescent material that is excited in the range of 350 to 400 nm (e.g. with UV-A light).

Images