Robust cell segmentation using high-PLEX protein and transcript images

Abstract

Described herein are systems and methods for robust and improved cell segmentation in a biological sample. Described herein are methods comprising applying an analysis machine-learning algorithm using high-plex protein and transcript images or data derived therefrom, and applying a machine-learning model for cell segmentation in the biological sample. Described herein are systems comprising a software module configured to apply an analysis machine-learning algorithm and a machine-learning model for cell segmentation in the biological sample.

Description

WSGR Docket No.68840-711.601 ROBUST CELL SEGMENTATION USING HIGH-PLEX PROTEIN AND TRANSCRIPT IMAGES CROSS-REFERENCE

[0001] This application claims the benefit of U.S. Provisional Application No.63 / 585,842, filed on September 27, 2023, which application is incorporated herein by reference in its entirety. BACKGROUND

[0002] In spatial “omics,” cell segmentation plays a pivotal role in ensuring the data quality of assays. Biological tissue samples can be taken from one or more subjects for analysis. The tissue samples can be used for diagnosis or analysis for determination of treatment strategies, or both. SUMMARY

[0003] Despite substantial progress in machine-learning-based cell segmentation, biological tissue samples taken from subjects present formidable challenges recognized by the Applicant. Applicant recognizes non-limiting issue of tissue processing and analysis such as biological tissue being difficult to accurately analyze due to the intrinsic characteristics of the tissue. These intrinsic characteristics may include densely packed cells, irregular cell shapes, and staining variations. Applicant recognizes that the high-plex spatial proteomics assays, while offering rich morphological data, have certain non-limiting challenges, including limited integration of information generated by the high-plex proteomics assays into cell segmentation models. For example, Applicant has recognized that channel-specific model architectures of pre-trained models have limited ability to integrate and process high-plex datasets, and necessitate costly and time-consuming retraining to address integration and processing of the high-plex datasets.

[0004] Applicant has recognized certain improvements directed toward these issues disclosed herein, such as a versatile framework designed for high-plex image-based cell segmentation. Applicant has recognized that this framework features flexibility in accommodating varying input-channel numbers, and harnesses pre-trained machine-learning models for cell segmentation without the need for time-consuming retraining. Applicant has recognized that high-plex protein datasets can be used to generate a pipeline addressing a wide variety of markers. Applicant has recognized that across all tested tissue types, this framework can recover cells with faint staining, accurately capturing intricate cell morphologies. Applicant has recognized that the modular design of the framework disclosed herein facilitates adaptationWSGR Docket No.68840-711.601 to new tissue types and novel sets of protein targets, promising a transformative future for cell biology research. The described framework recognized by Applicant, in some embodiments, is structured around two phases: analysis and application.

[0005] The analysis phase, in some embodiments, comprises randomly sampling the space to extract pixel-level high-plex protein expression profiles representative of the target tissue type. In some embodiments, correlation analysis can be performed on the extracted profiles to identify clusters of spatially colocalized proteins that serve as landmarks for specific cell types, or subcellular components, or tissue characteristics.

[0006] The application phase, in some embodiments, can comprise extending this grouping to new datasets of the same tissue type. In some embodiments, signals can be aggregated within each colocalization group. In some embodiments the aggregated signals can be provided as input to pre-trained machine-learning models, for example Cellpose. In some embodiments, group-specific cell segmentation information can be output, for example spatial flows and cell probabilities. In some embodiments, the output can be integrated from all groups to process the integrated spatial flows into the final cell mask.

[0007] In one aspect, disclosed herein are computer-implemented method for improving cell segmentation in a biological sample. In some embodiments, the method comprises (a) applying an analysis machine-learning algorithm for clustering, comprising: (i) obtaining a plurality of scanned images wherein each scanned image of the plurality of scanned images comprises one or more cells, wherein the one or more cells are labeled with a marker; (ii) extracting, by a computer processor, pixel-level expression data of the marker of each scanned image of the plurality of scanned images; (iii) determining, by the computer processor, a degree of correlation between the pixel level expression data and the spatial distribution of the marker of each scanned image of the plurality of scanned images; (iv) identifying, by the computer processor and based on the degree of correlation determined in (a)(iii), a plurality of clusters of the marker of each scanned image of the plurality of scanned images, wherein each cluster of the plurality of clusters comprises an identifier for a cell type, or a tissue characteristic, or subcellular components; (b) normalizing, by the computer processor the expression of each marker of each identifier of (a)(iv); (c) generating an aggregate intensity of expression of the markers for each cluster of the plurality of clusters identified in (a)(iv); (d) iteratively applying a segmentation machine-learning model to each cluster of the plurality of clusters identified in (a)(iv), comprising: (i) assessing spatial flows and cell probability of the cluster, based on the aggregated intensity of expression of (c); (ii) repeating the assessing for each cluster of the plurality of clusters; (e) combining, by the computer processor, data comprising the assessedWSGR Docket No.68840-711.601 spatial flows and cell probability of each cluster of the plurality of clusters; and (f) generating a cell segmentation mask based at least in part on the data of (e).

[0008] In some embodiments, the analysis machine-learning algorithm comprises an unsupervised machine-learning algorithm. In some embodiments, the analysis machine- learning algorithm comprises an unsupervised clustering algorithm. In some embodiments, the analysis machine-learning algorithm comprises one or more of a Louvian clustering algorithm, a Leiden clustering algorithm, or a K-means clustering algorithm, or any combination thereof.

[0009] In some embodiments, the segmentation machine-learning model comprises one or more types of segmentation machine-learning models. In some embodiments, the method further comprises applying a predetermined type of segmentation machine-learning model to each cluster of a subset of the plurality of clusters, wherein the subset of the plurality of clusters comprise the same identifier. In some embodiments, the method further comprises independently associating each identifier with a type of segmentation machine-learning model.

[0010] In some embodiments, the method comprises determining, by the computer processor, a cell boundary of each of the one or more cells of the biological sample. In some embodiments, the cell boundary is determined based at least in part on the combined data comprising the assessed spatial flows and the cell probability of each cluster of the plurality of clusters. In some embodiments, the method further comprises generating the cell segmentation mask based at least in part on the cell boundary of the one or more cells of the biological sample. In some embodiments, the method further comprises generating one or more cell labels based at least in part on the data of (d). In some embodiments, the method further comprises generating the cell segmentation mask based at least in part on the generated one or more cell labels.

[0011] In some embodiments, the computer-implemented method further comprises in (c), removing cells with small area or low confidence in segmentation prediction. In some embodiments, the computer-implemented method further comprises in (c), merging two different cells based on the area of intersection region. In some embodiments, the computer- implemented method further comprises in (c), splitting a cell based on the area of intersection region. In some embodiments, intensity of expression data comprises protein expression data, RNA expression data, or a combination thereof. In some embodiments, determining, in (a)(iii), a degree of correlation between the pixel level expression and the spatial distribution of the plurality of markers further comprises analyzing transcriptomic data, proteomic data, or both. In some embodiments, the transcriptomic data, proteomic data, or both comprise the human protein atlas database. In some embodiments, the computer-implemented method furtherWSGR Docket No.68840-711.601 comprises optimizing cell segmentation by applying a segmentation machine-learning model. The segmentation machine-learning model can utilize a Cellpose-style model, a Mesmer-style model, a StarDist model, or a deep watershed-based model.

[0012] In some embodiments, the plurality of markers comprises at least about 5, at least about 10, at least about 20, at least about 30, at least about 50, at least about 100, at least about 200 or at least about 300 different markers, including increments therein. In some embodiments, the marker comprises a subcellular component. In some embodiments, the subcellular component is a nucleus, a cell membrane, a cytoplasm, a soma, a protrusion, a mitochondria, a Golgi apparatus, a ribosome, an endoplasmic reticulum, a lysosome, a nucleolus, a nuclear speckles, a nucleoplasm, a nuclear periphery, a Cajal bodies, or a promyelocytic leukemia nuclear (PML) body. In some embodiments, the biological sample comprises a tissue sample comprising brain, prostate, lung, colon, breast, ovary, uterus, esophagus, stomach, liver, kidney, skin, heart, muscle, or bone marrow. In some embodiments, the plurality of cells comprises one or more cell types selected from a neuron, an astrocyte, a microglial, an immune cell, an endothelial cell, an epithelial cell, a hepatocyte, a keratinocyte, a myoepithelial cell, or a muscle cell.

[0013] In some embodiments, the method further comprises generating, by the computer processor, one or more raw cell probability images. In some embodiments, the method further comprises assigning a value of zero to each existing cell label of the plurality of markers. In some embodiments, the method further comprises identifying one or more combined cell clusters. In some embodiments, the method further comprises applying a threshold value for assessing spatial flows and cell probability of the one or more combined cell clusters. In some embodiments, the threshold value is (1x10-1, max), (1x10-2, max), (1x10-3, max), (1x10-4, max), (1x10-5, max), (1x10-6, max), (1x10-7, max), (1x10-8, max), (1x10-9, max), (1x10-10, max), (1x10-11, max), (1x10-12, max), (1x10-13, max), (1x10-14, max), (1x10-15, max), or less than (1x10-15, max). In some embodiments, the threshold value is (1x10-5, max).

[0014] In some embodiments, the method further comprises applying a size-range filter using the threshold value. In some embodiments, the method further comprises generating one or more new cell labels for the one or more combined cell clusters. In some embodiments, the one or more new cell labels comprise a marker. In some embodiments, identifying one or more combined cell clusters comprises setting a threshold value of (0, max) or (-2, max) for assessing spatial flows and cellular probability of the one or more combined clusters.WSGR Docket No.68840-711.601

[0015] In some embodiments, the method further comprises applying a minimum size filter to each combined cluster of the one or more combined clusters. In some embodiments, generating the one or more new cell labels further comprises reassigning one or more of the existing cell labels. In some embodiments, reassigning the one or more labels comprises determining the intersection of area (IoA) value of each combined cluster of the one or more combined clusters. In some embodiments, the IoA comprises the touch IoA of each combined cluster of the one or more combined clusters. In some embodiments, the method further comprises removing one or more of the labels from each combined cluster of the one or more combined clusters having an IoA threshold value of less than 0.6. In some embodiments, the method further comprises selecting one or more of the existing labels to reassign by applying a nearest neighbor split algorithm. In some embodiments, the nearest neighbor split algorithm is performed against the existing cell labels and the new cell labels. In some embodiments, the nearest neighbor split algorithm is performed to generate one or more newly detected cell borders. In some embodiments, the method further comprises generating the one or more new cell labels for one or more newly detected cells comprising the one or more newly detected cell borders. In some embodiments, the method further comprises generating final cell labels by combining the new cell labels and the existing cell labels. In some embodiments, the method further comprises generating a segmentation mask comprising the final cell labels.

[0016] In another aspect, described herein are systems comprising one or more computer processors and computer memory coupled thereto, the computer memory comprising a machine executable code that, upon execution by the one or more computer processors, implements the method of described herein.

[0017] In another aspect, disclosed herein are computer-implemented systems comprising a computing device comprising at least one processor and instructions executable by the at least one processor to provide an application for improved cell segmentation in a biological sample. In some embodiments, the system comprises a) a software module configured to apply an analysis machine-learning algorithm for clustering, the analysis machine-learning algorithm configured to perform operations comprising: (i) obtaining a plurality of scanned images, wherein each scanned image of the plurality of scanned images comprises one or more cells, wherein the one or more cells are labeled with a marker; (ii) extracting, by a computer processor, pixel-level expression data of the marker of each scanned image of the plurality of scanned images; (iii) determining, by the computer processor, a degree of correlation between the pixel level expression data and the spatial distribution of the marker of each scanned image of the plurality of scanned images; (iv) identifying, by the computer processor and based onWSGR Docket No.68840-711.601 the degree of correlation determined in (a)(iii), a plurality of clusters of the marker of each scanned image of the plurality of scanned images, wherein each cluster of the plurality of clusters comprises an identifier for a cell type, or subcellular components, or a tissue characteristic; (b) a software module configured to normalize the expression of each marker of each identifier of (a)(iv); (c) a software module configured to generate an aggregate intensity of expression of the markers for each cluster of the plurality of clusters identified in (a)(iv); (d) a software module configured to iteratively apply a segmentation machine-learning algorithm to each cluster of the plurality of clusters identified in (a)(iv), comprising: (i) assessing spatial flows and cell probability of the cluster, based on the aggregated intensity of expression of (c); (ii) repeating the assessing for each cluster of the plurality of clusters; (e) a software module configured to combine data comprising the assessed spatial flows and cell probability of each cluster of the plurality of clusters; and (f) a software module configured to perform operations comprising generating a cell segmentation mask based at least in part on the data of (e).

[0018] In some embodiments, the analysis machine-learning algorithm comprises an unsupervised machine-learning algorithm. In some embodiments, the analysis machine- learning algorithm comprises an unsupervised clustering algorithm. In some embodiments, the analysis machine-learning algorithm comprises one or more of a Louvian clustering algorithm, a Leiden clustering algorithm, or a K-means clustering algorithm, or any combination thereof. In some embodiments, the segmentation machine-learning model comprises one or more types of segmentation machine-learning models.

[0019] In some embodiments, the operations further comprise applying a predetermined type of segmentation machine-learning model to each cluster of a subset of the plurality of clusters, wherein the subset of the plurality of clusters comprise the same identifier. In some embodiments, the operations further comprise independently associating each identifier with a type of segmentation machine-learning model.

[0020] In some embodiments, the software module configured to apply the segmentation machine-learning model in (c) is further configured to perform operations comprising determining a cell boundary of each of the one or more cells of the biological sample. In some embodiments, the software module configured to apply the segmentation machine-learning model in (c) is further configured to perform operations comprising determining a cell boundary based at least in part on the combined data comprising the assessed spatial flows and the cell probability of each cluster of the plurality of clusters. In some embodiments, the software module configured to apply the segmentation machine-learning model in (c) is further configured to perform operations comprising generating the cell segmentation mask based atWSGR Docket No.68840-711.601 least in part on the cell boundary of the one or more cells of the biological sample. In some embodiments, the software module configured to apply the segmentation machine-learning model in (c) is further configured to perform operations comprising generating one or more cell labels based at least in part on the data of (d). In some embodiments, the software module configured to apply the segmentation machine-learning model in (c) is further configured to perform operations comprising generating the cell segmentation mask based at least in part on the generated one or more cell labels.

[0021] In some embodiments, the software module configured to apply the segmentation machine-learning model in (c) further comprises removing cells with small area or low confidence in segmentation prediction. In some embodiments, the software module configured to apply the segmentation machine-learning model in (c) further comprises merging two different cells based on the area of intersection region. In some embodiments, the software module configured to apply the segmentation machine-learning model in (c) further comprises splitting a cell based on the area of intersection region. In some embodiments, pixel level expression data comprises protein expression data, RNA expression data, or a combination thereof. In some embodiments, determining, in (a)(iii), a degree of correlation between the pixel level expression and the spatial distribution of the plurality of markers further comprises analyzing transcriptomic data, proteomic data, or both. In some embodiments, the transcriptomic data, proteomic data, or both comprise the human protein atlas database. In some embodiments, the computer-implemented system further comprises optimizing cell segmentation by applying a segmentation machine-learning model. The segmentation machine-learning model can utilize a Cellpose-style model, a Mesmer-style model, a StarDist model, or a deep watershed-based model.

[0022] In some embodiments, the plurality of markers comprises at least about 5, at least about 10, at least about 20, at least about 30, at least about 50, at least about 100, at least about 200 or at least about 300 different markers, including increments therein. In some embodiments, the cell marker can be a subcellular component. In some embodiments, the subcellular component is a nucleus, a cell membrane, a cytoplasm, a soma, a protrusion, a mitochondria, a Golgi apparatus, a ribosome, an endoplasmic reticulum, a lysosome, a nucleolus, a nuclear speckles, a nucleoplasm, a nuclear periphery, a Cajal bodies, or a promyelocytic leukemia nuclear (PML) body. In some embodiments, the biological sample comprises a tissue type comprising brain, prostate, lung, colon, breast, ovary, uterus, esophagus, stomach, liver, kidney, skin, heart, muscle, or bone marrow. In some embodiments, the plurality of cells comprises one or more cell types selected from a neuron, an astrocyte, a microglial, an immuneWSGR Docket No.68840-711.601 cell, an endothelial cell, an epithelial cell, a hepatocyte, a keratinocyte, a myoepithelial cell, or a muscle cell.

[0023] In some embodiments the application further comprises generating, by the computer processor, one or more raw cell probability images. In some embodiments the application further comprises assigning a value of zero to each existing cell label of the plurality of markers. In some embodiments the application further comprises identifying one or more combined cell clusters.

[0024] In some embodiments, the operations further comprise applying a threshold value for assessing spatial flows and cell probability of the one or more combined cell clusters. In some embodiments, the threshold value is (1x10-1, max), (1x10-2, max), (1x10-3, max), (1x10-4, max), (1x10-5, max), (1x10-6, max), (1x10-7, max), (1x10-8, max), (1x10-9, max), (1x10-10, max), (1x10-11, max), (1x10-12, max), (1x10-13, max), (1x10-14, max), (1x10-15, max), or less than (1x10-15, max). In some embodiments, the threshold value is (1x10-5, max).

[0025] In some embodiments the application further comprises applying a size-range filter using the threshold value. In some embodiments the application further comprises generating one or more new cell labels for the one or more combined cell clusters. In some embodiments, the one or more new cell labels comprise a marker. In some embodiments, identifying one or more combined cell clusters comprises setting a threshold value of (0, max) or (-2, max) for assessing spatial flows and cellular probability of the one or more combined clusters.

[0026] In some embodiments the application further comprises applying a minimum size filter to each combined cluster of the one or more combined clusters. In some embodiments the application further comprises generating the one or more new cell labels further comprises reassigning one or more of the existing cell labels. In some embodiments, reassigning the one or more labels comprises determining the intersection of area (IoA) value of each combined cluster of the one or more combined clusters. In some embodiments, the IoA comprises the touch IoA of each combined cluster of the one or more combined clusters. In some embodiments the application further comprises removing one or more of the labels from each combined cluster of the one or more combined clusters having an IoA threshold value of less than 0.6.

[0027] In some embodiments the application further comprises selecting one or more of the existing labels to reassign by applying a nearest neighbor split algorithm. In some embodiments, the nearest neighbor split algorithm is performed against the existing cell labelsWSGR Docket No.68840-711.601 and the new cell labels. In some embodiments, the nearest neighbor split algorithm is performed to generate one or more newly detected cell borders.

[0028] In some embodiments the application further comprises generating the one or more new cell labels for one or more newly detected cells comprising the one or more newly detected cell borders. In some embodiments the application further comprises generating final cell labels by combining the new cell labels and the existing cell labels. In some embodiments the application further comprises generating a segmentation mask comprising the final cell labels.

[0029] In yet another embodiment, disclosed herein is a computer-implemented method for improving cell segmentation in a biological sample, the method comprising: (a) receiving grouping data by the computer processor, the grouping data comprising -a plurality of previously identified clusters of markers, where each previously identified cluster comprises a unique spatial expression pattern of the markers; (b) normalizing by the computer processor, the expression of each marker of each previously identified cluster; (c) generating an aggregate intensity of expression of the markers for each cluster of the plurality of clusters identified in (a)(iv); (d) iteratively applying a segmentation machine-learning model to each cluster of the previously identified clusters of markers, wherein the machine-learning model is configured to perform operations comprising: (i) assessing spatial flows and cell probability of the cluster, based on the aggregate intensity of expression of (c), (iii) repeating the assessing for each cluster of the plurality of previously identified clusters; (e) combining, by the computer processor, data comprising the assessed spatial flows and cell probability of each cluster of the plurality of previously identified clusters; and (f) generating a cell segmentation mask based at least in part on the data of (e).

[0030] In yet another embodiment, described herein is a computer-implemented system comprising a computing device comprising at least one processor and instructions executable by the at least one processor to provide an application for improved cell segmentation in a biological sample, comprising: (a) a software module configured to receive grouping data comprising a plurality of previously identified clusters of markers, where each previously identified cluster comprises a unique spatial expression pattern of the markers; (b) a software module configured to normalize the expression of each marker of each previously identified cluster; (c) a software module configured to generate an aggregate intensity of expression of the markers for each cluster of the plurality of clusters identified in (a)(iv); (d) a software module configured to iteratively apply a segmentation machine-learning model to each cluster of the plurality of previously identified clusters of markers, wherein the machine-learning model is configured to perform operations comprising: (i) assessing spatial flows and cellWSGR Docket No.68840-711.601 probability of the cluster, based on the aggregate intensity of expression of (c), (ii) repeating the assessing for each cluster of the plurality of previously identified clusters; (e) a software module configured to combine data comprising the assessed spatial flows and cell probability of each cluster of the plurality of previously identified clusters; and (f) a software module configured to perform operations comprising generating a cell segmentation mask based at least in part on the data of (e).

[0031] In some embodiments, the method further comprises generating, by the computer processor, one or more raw cell probability images. In some embodiments, the method further comprises assigning a value of zero to each existing cell label of the plurality of markers. In some embodiments, the method further comprises identifying one or more combined cell clusters. In some embodiments, the method further comprises applying a threshold value of (1x10-5, max) for assessing spatial flows and cell probability of the one or more combined cell clusters. In some embodiments, the method further comprises applying a size-range filter using the threshold value. In some embodiments, the method further comprises further comprising generating one or more new cell labels for the one or more combined cell clusters. In some embodiments, the one or more new cell labels comprise a marker.

[0032] In some embodiments, the method further comprises identifying one or more combined cell clusters comprises setting a threshold value of (0, max) or (-2, max) for assessing spatial flows and cellular probability of the one or more combined clusters. In some embodiments, the method further comprises applying a minimum size filter to each combined cluster of the one or more combined clusters. In some embodiments, generating the one or more new cell labels further comprises reassigning one or more of the existing cell labels. In some embodiments, reassigning the one or more labels comprises determining the intersection of area (IoA) value of each combined cluster of the one or more combined clusters. In some embodiments, the IoA comprises the touch IoA of each combined cluster of the one or more combined clusters.

[0033] In some embodiments, the method further comprises comprising removing one or more of the labels from each combined cluster of the one or more combined clusters having an IoA threshold value of less than 0.6. In some embodiments, the method further comprises selecting one or more of the existing labels to reassign by applying a nearest neighbor split algorithm. In some embodiments, the nearest neighbor split algorithm is performed against the existing cell labels and the new cell labels. In some embodiments, the nearest neighbor split algorithm is performed to generate one or more newly detected cell borders. In some embodiments, the method further comprises generating the one or more new cell labels for one or more newly detected cells comprising the one or more newly detected cell borders. In some embodiments,WSGR Docket No.68840-711.601 the method further comprises generating final cell labels by combining the new cell labels and the existing cell labels. In some embodiments, the method further comprises generating a segmentation mask comprising the final cell labels. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] A better understanding of the features and advantages of the present subject matter will be obtained by reference to the following detailed description that sets forth illustrative embodiments and the accompanying drawings of which:

[0035] FIG.1 shows a non-limiting example of a computing device; in this case, a device with one or more processors, memory, storage, and a network interface.

[0036] FIG.2 illustrates a non-limiting example of a process flow for a framework described herein for cell segmentation using high-plex protein and transcript images; in this case, a process flow with associated mathematical formula.

[0037] FIG.3 shows a non-limiting example of an intersection of union analysis as part of an optional post-processing step.

[0038] FIG.4 illustrates a non-limiting example of an optional post-processing step, in this case an optional post-processing step for splitting of cells by nuclei.

[0039] FIGS.5A and 5B illustrate an optional post-processing step, in this case an optional post-processing step for capturing long processes.

[0040] FIG.6 shows non-limiting example results of validation experiments, in this case a single merge imaging and a second merge imaging.

[0041] FIG.7 shows non-limiting example results of validation experiments, in this case in adjacent normal tissue and malignant tissue.

[0042] FIGS.8A, 8B, and 8C depict non-limiting examples of differential position (dP) spatial gradient flows input, predicted dP flows, and resulting predicted cell labels based on converting the dP flows.

[0043] FIGS.9A and 9B illustrate non-limiting examples of various isolated dP flow levels in default mask conversion, where recovery of cell labels is incomplete.

[0044] FIG.10 shows non-limiting examples of various mask layers at various cell probability and flow cutoff combinations, where recovery of cell labels is incomplete.

[0045] FIG.11 illustrates a non-limiting exemplary workflow for 2-step mask generation.WSGR Docket No.68840-711.601

[0046] FIG.12 illustrates a non-limiting example of mask and imaging outputs at various steps in an exemplary 2-step mask generation method.

[0047] FIG.13 depicts non-limiting examples of applying 2-step mask generation methods for an exemplary selection of input imaging data and intermediate outputs at various steps of the 2- step mask generation method.

[0048] FIG.14 depicts non-limiting examples of error correction performed on exemplary samples of large size range and highly irregular cell shapes using an exemplary masking method.

[0049] FIG.15 depicts non-limiting examples of masks for exemplary dense nuclei samples.

[0050] FIG.16 shows non-limiting examples of recovery of weak-flow cell mask imaging using the exemplary 2-step mask method.

[0051] FIG.17 illustrates non-limiting examples of recovery of weak-flow nuclei cell mask imaging using the exemplary 2-step mask method.

[0052] FIG.18 depicts a non-limiting example of a multi-protein 2-step mask generation exemplary workflow.

[0053] FIG.19 illustrates a non-limiting example of a unified workflow for an exemplary 2-step mask generation method workflow.

[0054] FIG.20A illustrates an exemplary tissue-type based unified workflow for an exemplary 2-step mask generation method for various tissues.

[0055] FIG.20B depicts a non-limiting example of a labeling and mask combination method for various exemplary thresholds of intersection of union (IoU) and intersection of area (IoA) values of cells.

[0056] FIG.21 depicts a non-limiting example of a workflow from pre-identified data input for a framework described herein for cell segmentation using high-plex protein and transcript images; in this case, a process flow with associated mathematical formulas.WSGR Docket No.68840-711.601 DETAILED DESCRIPTION

[0057] Described herein, in certain embodiments, are computer-implemented methods for improving cell segmentation in a biological sample.

[0058] In one aspect, disclosed herein are computer-implemented method for improving cell segmentation in a biological sample. In one aspect, disclosed herein can be computer- implemented method for improving cell segmentation in a biological sample. In some embodiments, the method can comprise (a) applying an analysis machine-learning algorithm for clustering, which can comprise: (i) obtaining a plurality of scanned images wherein each scanned image of the plurality of scanned images can comprise one or more cells, wherein the one or more cells can be labeled with a marker. In some embodiments, the method can further comprise (ii) extracting, by a computer processor, pixel-level expression data of the marker of each scanned image of the plurality of scanned images. In some embodiments, the method can further comprise (iii) determining, by the computer processor, a degree of correlation between the aggregate intensity of expression data and the spatial distribution of the marker of each scanned image of the plurality of scanned images. In some embodiments, the method can further comprise (iv) identifying, by the computer processor and based on the degree of correlation determined in (a)(iii), a plurality of clusters of the marker of each scanned image of the plurality of scanned images, wherein each cluster of the plurality of clusters can comprise an identifier for a cell type, or a subcellular component, or a tissue characteristic. In some embodiments, the method can further comprise (b) normalizing, by the computer processor, the expression of each marker of each identifier of (a)(iv). In some embodiments, the method can further comprise (c) generating an aggregate intensity of expression of the markers for each cluster of the plurality of clusters identified in (a)(iv). In some embodiments, the method can further comprise (d) iteratively applying a segmentation machine-learning model to each cluster of the plurality of clusters identified in (a)(iv). In some embodiments, the method can further comprise (i) assessing spatial flows and cell probability of the cluster, based on the combined pixel level expression of (c). In some embodiments, the method can further comprise (ii) repeating the assessing for each cluster of the plurality of clusters. In some embodiments, the method can further comprise (e) combining, by the computer processor, data which can comprise the assessed spatial flows and cell probability of each cluster of the plurality of clusters. In some embodiments, the method can further comprise and (f) generating a cell segmentation mask based at least in part on the data of (e).

[0059] In some embodiments, the analysis machine-learning algorithm can comprise an unsupervised machine-learning algorithm. In some embodiments, the analysis machine-learningWSGR Docket No.68840-711.601 algorithm can comprise an unsupervised clustering algorithm. In some embodiments, the analysis machine-learning algorithm can comprise one or more of a Louvian clustering algorithm, a Leiden clustering algorithm, or a K-means clustering algorithm, or any combination thereof. In some embodiments, the analysis machine-learning algorithm can comprise one or more of a distribution-based clustering algorithm, a centroid clustering algorithm, a density- based clustering algorithm, or another type of clustering algorithm. In some embodiments, the analysis machine-learning algorithm can comprise one or more of an affinity propagation algorithm, an agglomerative hierarchical clustering algorithm, a balanced iterative reducing and clustering algorithm (BIRCH), a density-based spatial clustering algorithm (DBSCAN), a Gaussian mixture algorithm, a mean shift clustering algorithm, a mini-batch k-means algorithm, an OPTICS algorithm, a spectral clustering algorithm, or any combination thereof.

[0060] In some embodiments, the segmentation machine-learning model can comprise one or more types of segmentation machine-learning models. In some embodiments, the method can further comprise applying a predetermined type of segmentation machine-learning model to each cluster of a subset of the plurality of clusters. In some embodiments, the subset of the plurality of clusters can comprise the same identifier. In some embodiments, the method can further comprise independently associating each identifier with a type of segmentation machine- learning model. In some embodiments, the segmentation machine-learning model can comprise MaskRCNN, CenterMask, YOLO, LOIs, location proposal networks, region proposal networks, LACSS, or any combination thereof.

[0061] In some embodiments, the method can comprise determining, by the computer processor, a cell boundary of each of the one or more cells of the biological sample. In some embodiments, the cell boundary can be determined based at least in part on the combined data which can comprise the assessed spatial flows and the cell probability of each cluster of the plurality of clusters. In some embodiments, the method further can comprise generating the cell segmentation mask based at least in part on the cell boundary of the one or more cells of the biological sample. In some embodiments, the method further can comprise generating one or more cell labels based at least in part on the data of (d). In some embodiments, the method further comprises generating the cell segmentation mask based at least in part on the generated one or more cell labels.

[0062] In some embodiments, the method can further comprise generating, by the computer processor, one or more raw cell probability images. In some embodiments, the method can further comprise assigning a value of zero to each existing cell label of the plurality of markers. In some embodiments, the method can further comprise identifying one or more combined cellWSGR Docket No.68840-711.601 clusters. In some embodiments, the method can further comprise applying a threshold value for assessing spatial flows and cell probability of the one or more combined cell clusters. In some embodiments, the threshold value can be (1x10-1, max), (1x10-2, max), (1x10-3, max), (1x10-4, max), (1x10-5, max), (1x10-6, max), (1x10-7, max), (1x10-8, max), (1x10-9, max), (1x10-10, max), (1x10-11, max), (1x10-12, max), (1x10-13, max), (1x10-14, max), (1x10-15, max), or less than (1x10-15, max). In some embodiments, the threshold value can be (1x10-5, max).

[0063] In some embodiments, the method can further comprise applying a size-range filter using the threshold value. In some embodiments, the method can further comprise generating one or more new cell labels for the one or more combined cell clusters. In some embodiments, the one or more new cell labels can comprise a marker. In some embodiments, identifying one or more combined cell clusters can comprise setting a threshold value of (0, max) or (-2, max) for assessing spatial flows and cellular probability of the one or more combined clusters.

[0064] In some embodiments, the method can further comprise applying a minimum size filter to each combined cluster of the one or more combined clusters. In some embodiments, generating the one or more new cell labels can further comprise reassigning one or more of the existing cell labels. In some embodiments, reassigning the one or more labels can comprise determining the intersection of area (IoA) value of each combined cluster of the one or more combined clusters. In some embodiments, the IoA can comprise the touch IoA of each combined cluster of the one or more combined clusters. In some embodiments, the method can further comprise removing one or more of the labels from each combined cluster of the one or more combined clusters which can have an IoA threshold value of less than 0.6. In some embodiments, the method can further comprise selecting one or more of the existing labels to reassign by applying a nearest neighbor split algorithm. In some embodiments, the nearest neighbor split algorithm can be performed against the existing cell labels and the new cell labels. In some embodiments, the nearest neighbor split algorithm can be performed to generate one or more newly detected cell borders. In some embodiments, the method can further comprise generating the one or more new cell labels for one or more newly detected cells which can comprise the one or more newly detected cell borders. In some embodiments, the method can further comprise generating final cell labels by combining the new cell labels and the existing cell labels. In some embodiments, the method can further comprise generating a segmentation mask which can comprise the final cell labels.

[0065] Also described herein, in certain embodiments, are computer-implemented systems comprising a computing device comprising at least one processor and instructions executable by the at least one processor to provide an application for improved cell segmentation in aWSGR Docket No.68840-711.601 biological sample. In some embodiments, the system comprises a) a software module configured to apply an analysis machine-learning algorithm for clustering, the analysis machine-learning algorithm configured to perform operations comprising: (i) obtaining a plurality of scanned images, wherein each scanned image of the plurality of scanned images comprises one or more cells, wherein the one or more cells are labeled with a marker; (ii) extracting, by a computer processor, pixel-level expression data of the marker of each scanned image of the plurality of scanned images; (iii) determining, by the computer processor, a degree of correlation between the pixel level expression data and the spatial distribution of the marker of each scanned image of the plurality of scanned images; (iv) identifying, by the computer processor and based on the degree of correlation determined in (a)(iii), a plurality of clusters of the marker of each scanned image of the plurality of scanned images, wherein each cluster of the plurality of clusters comprises an identifier for a cell type, or a subcellular component, or a tissue characteristic; (b) a software module configured to normalize the expression of each marker of each identifier of (a)(iv); (c) a software module configured to generate an aggregate intensity of expression of the markers for each cluster of the plurality of clusters identified in (a)(iv); (d) a software module configured to iteratively apply a segmentation machine-learning algorithm to each cluster of the plurality of clusters identified in (a)(iv), comprising: (i) assessing spatial flows and cell probability of the cluster, based on the combined intensity of expression of (c); (ii) repeating the assessing for each cluster of the plurality of clusters; (e) a software module configured to combine data comprising the assessed spatial flows and cell probability of each cluster of the plurality of clusters; and (f) a software module configured to perform operations comprising generating a cell segmentation mask based at least in part on the data of (e). Certain definitions

[0066] Unless otherwise defined, all technical terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which the present subject matter belongs.

[0067] As used in this specification and the appended claims, the singular forms “a,” “an,” and “the” include plural references unless the context clearly dictates otherwise. Any reference to “or” herein is intended to encompass “and / or” unless otherwise stated.

[0068] Reference throughout this specification to “some embodiments,” “further embodiments,” or “a particular embodiment,” means that a particular feature, structure, or characteristic described in connection with the embodiment is included in at least one embodiment. Thus, the appearances of the phrase “in some embodiments,” or “in further embodiments,” or “in aWSGR Docket No.68840-711.601 particular embodiment” in various places throughout this specification are not necessarily all referring to the same embodiment. Furthermore, the particular features, structures, or characteristics may be combined in any suitable manner in one or more embodiments.

[0069] As used herein, the term “about” in some cases refers to an amount that is approximately the stated amount, in some cases near the stated amount by 10%, 5%, or 1%, including increments therein, and in some cases, in reference to a percentage, refers to an amount that is greater or less the stated percentage by 10%, 5%, or 1%, including increments therein.

[0070] As used herein, the phrases “at least one,” “one or more,” and “and / or” are open-ended expressions that are both conjunctive and disjunctive in operation. For example, each of the expressions “at least one of A, B and C,” “at least one of A, B, or C,” “one or more of A, B, and C”, “one or more of A, B, or C” and “A, B, and / or C” means A alone, B alone, C alone, A and B together, A and C together, B and C together, or A, B and C together. As used herein, the phrase “at most three” can mean less than one, one, two, or three. Cell Segmentation Methods

[0071] In one non-limiting embodiment, the method can comprise a computer-implemented method for improving cell segmentation in a biological sample comprising applying an analysis machine-learning algorithm for clustering. In some embodiments, an analysis machine-learning algorithm can be configured to perform operations comprising obtaining a plurality of scanned images. In some embodiments, each of the plurality of scanned images can comprise a plurality of cells. In some embodiments, the plurality of cells can be labeled. In some embodiments, the plurality of cells can be labeled with a plurality of markers. In some embodiments, the plurality of markers can comprise one or more of fluorescent markers, antibody markers, nanoparticle markers, probes, stains, genetic tags, membrane dyes, biosensors, or any combination thereof.

[0072] In some embodiments, the method can further comprise extracting, by a computer processor, pixel-level expression data of the plurality of markers. In some embodiments, the method can further comprise determining, by the computer processor, a degree of correlation between the pixel level expression and the spatial distribution of the plurality of markers on each of the plurality of scanned images. In some embodiments, the method can further comprise identifying, by the computer processor and based on the degree of correlation, a plurality of clusters of at least one subset of the plurality of markers. In some embodiments, each of the plurality of clusters can comprise an identifier for a cell type, or a tissue characteristic, or a subcellular component.WSGR Docket No.68840-711.601

[0073] In some embodiments, the method can further comprise applying the analysis machine- learning algorithm to segment cells in the biological sample. In some embodiments, the machine learning algorithm to segment cells in the biological sample can comprise normalizing and aggregating, by the computer processor, the pixel level expression of the plurality of clusters of at least one subset of the plurality of markers.

[0074] In some embodiments, the method can further comprise analyzing, by the computer processor, the aggregate intensity of expression of each of the plurality of clusters. In some embodiments, the method can further comprise assessing spatial flows of each of the plurality of clusters. In some embodiments, the method can further comprise assessing cell probability of each of the plurality of clusters. In some embodiments, the method can further comprise assessing spatial flows and cell probability of each of the plurality of clusters. In some embodiments, the method can assess spatial flows and cell probability applying the analysis machine-learning algorithm based on the aggregate intensity of expression.

[0075] In some embodiments, the method can further comprise determining a cell boundary of each of the plurality of cells in the biological sample. In some embodiments, determining the cell boundary can be performed by the computer processor. In some embodiments, determining the cell boundary can be based on the combined spatial flows and cell probability of each of the plurality of clusters.

[0076] In some embodiments, the method can further comprise removing cells with small area or low confidence in segmentation prediction. In some embodiments, the method can further comprise merging two different cells. In some embodiments, the two different cells can be merged based on the area of intersection region. In some embodiments, the method can further comprise splitting a cell based on the area of intersection region.

[0077] In some embodiments, intensity of expression data can comprise protein expression data, RNA expression data, or a combination thereof. In some embodiments, the method can further comprise determining a degree of correlation between the intensity of expression and the spatial distribution of the plurality of markers. In some embodiments, determining the degree of correlation can further comprise analyzing transcriptomic data, proteomic data, or both. In some embodiments, the transcriptomic data, proteomic data, or both can further comprise the human protein atlas database. In some embodiments, the transcriptomic data, proteomic data, or both can comprise one or more unique gene expression profiles. In some embodiments, the transcriptomic data, proteomic data, or both can comprise one or more epigenetic profiles.WSGR Docket No.68840-711.601

[0078] In some embodiments, the method can further comprise optimizing cell segmentation by applying a segmentation machine-learning model. In some embodiments, the segmentation machine-learning model can utilize a Cellpose-style model, a Mesmer-style model, a StarDist model, or a deep watershed-based model. In some embodiments, the plurality of markers can comprise at least about 5, at least about 10, at least about 20, at least about 30, at least about 50, at least about 100, at least about 200 or at least about 300 different markers. In some embodiments, the plurality of markers can comprise at least about 2, at least about 3, at least about 4, at least about 5, at least about 6, at least about 7, at least about 8, at least about 9, at least about 10, at least about 11, at least about 12, at least about 13, at least about 14, at least about 15, at least about 16, at least about 17, at least about 18, at least about 19, at least about 20, at least about 21, at least about 22, at least about 23, at least about 24, at least about 25, at least about 26, at least about 27, at least about 28, at least about 29, at least about 30, at least about 31, at least about 32, at least about 33, at least about 34, at least about 35, at least about 36, at least about 37, at least about 38, at least about 39, at least about 40, at least about 41, at least about 42, at least about 43, at least about 44, at least about 45, at least about 46, at least about 47, at least about 48, at least about 49, at least about 50, at least about 51, at least about 52, at least about 53, at least about 54, at least about 55, at least about 56, at least about 57, at least about 58, at least about 59, at least about 60, at least about 61, at least about 62, at least about 63, at least about 64, at least about 65, at least about 66, at least about 67, at least about 68, at least about 69, at least about 70, at least about 71, at least about 72, at least about 73, at least about 74, at least about 75, at least about 76, at least about 77, at least about 78, at least about 79, at least about 80, at least about 81, at least about 82, at least about 83, at least about 84, at least about 85, at least about 86, at least about 87, at least about 88, at least about 89, at least about 90, at least about 91, at least about 92, at least about 93, at least about 94, at least about 95, at least about 96, at least about 97, at least about 98, at least about 99, or at least about 100 different markers. In some embodiments, the plurality of markers can comprise at least about 100, at least about 110, at least about 120, at least about 130, at least about 140, at least about 150, at least about 160, at least about 170, at least about 180, at least about 190, at least about 200, at least about 210, at least about 220, at least about 230, at least about 240, at least about 250, at least about 260, at least about 270, at least about 280, at least about 290, at least about 300, at least about 310, at least about 320, at least about 330, at least about 340, at least about 350, at least about 360, at least about 370, at least about 380, at least about 390, at least about 400, at least about 410, at least about 420, at least about 430, at least about 440, at least about 450, at least about 460, at least about 470, at least about 480, at least about 490, or at least about 500 different markers, including increments therein.WSGR Docket No.68840-711.601

[0079] In some embodiments, the subcellular component can comprise a nucleus, a cell membrane, a cytoplasm, a soma, a protrusion, a mitochondria, a Golgi apparatus, a ribosome, an endoplasmic reticulum, a lysosome, a nucleolus, a nuclear speckles, a nucleoplasm, a nuclear periphery, a Cajal bodies, or a promyelocytic leukemia nuclear (PML) body.

[0080] In some embodiments, the tissue type can comprise brain, prostate, lung, colon, breast, ovary, uterus, esophagus, stomach, liver, kidney, skin, heart, muscle, or bone marrow. In some embodiments, the tissue type can comprise epithelial tissue, connective tissue, muscle tissue, nervous tissue, blood tissue, bone tissue, adipose tissue, cartilage tissue, cardiac muscle tissue, skeletal muscle tissue, smooth muscle tissue, lymphatic tissue, vascular tissue, or any combination thereof.

[0081] In some embodiments, the plurality of cells can comprise one or more cell types selected from a neuron, an astrocyte, a microglial, an immune cell, an endothelial cell, an epithelial cell, a hepatocyte, a keratinocyte, a myoepithelial cell, or a muscle cell. In some embodiments, the one or more cell types can comprise neuronal cells, muscle cells, adipocytes, hepatocytes, leukocytes, erythrocytes, osteocytes, fibroblasts, keratinocytes, endothelial cells, lymphocytes, phagocytes, cardiomyocytes, nephrons, alveolar cells, enterocytes, astrocytes, oocytes, spermatocytes, zygote, stem cells, induced pluripotent stem cells, epithelial cells, pancreatic cells, retinal cells, hair cells, or any combination thereof.

[0082] In another embodiment, disclosed herein is a system comprising one or more computer processors and computer memory coupled thereto, the computer memory comprising a machine executable code that, upon execution by the one or more computer processors, implements the method described herein.

[0083] In some embodiments, the system can comprise a software module that can be configured to apply an analysis machine-learning algorithm for clustering. In some embodiments, the analysis machine-learning algorithm can be configured to perform operations that can comprise obtaining a plurality of scanned images. In some embodiments, each scanned image of the plurality of scanned images can comprise one or more cells. In some embodiments, the analysis machine-learning algorithm can be configured to perform operations that can comprise labeling the one or more cells with a marker. In some embodiments, the analysis machine-learning algorithm can be configured to perform operations that can comprise extracting, pixel-level expression data of the marker of each scanned image of the plurality of scanned images. In some embodiments, the analysis machine-learning algorithm can be configured to perform operations that can comprise determining a degree of correlation betweenWSGR Docket No.68840-711.601 the pixel level expression data and the spatial distribution of the marker of each scanned image of the plurality of scanned images. In some embodiments, the analysis machine-learning algorithm can be configured to perform operations that can comprise normalizing the expression of each marker of the plurality of scanned images. In some embodiments, each cluster of the plurality of clusters can comprise an identifier for a cell type, or a subcellular component, or a cellular subtype, or a tissue characteristic. In some embodiments, a software module can be configured to iteratively apply a segmentation machine-learning model to each cluster of the plurality of clusters. In some embodiments, the segmentation machine-learning model can be configured to perform operations that can comprise normalizing and aggregating the pixel level expression of the marker of the cluster. In some embodiments, the segmentation machine- learning model can be configured to perform operations that can comprise analyzing the combined pixel level expression of the cluster. In some embodiments, the segmentation machine-learning model can be configured to perform operations that can comprise assessing spatial flows and cell probability of the cluster, based on the combined pixel level expression. In some embodiments, the segmentation machine-learning model can be configured to perform operations that can comprise repeating the operations for each cluster of the plurality of clusters. In some embodiments, a software module can be configured to combine data can comprise the assessed spatial flows and cell probability of each cluster of the plurality of clusters. In some embodiments, a software module can be configured to perform operations can comprise generating a cell segmentation mask based at least in part on the data comprising the assessed spatial flows and cell probability of each cluster of the plurality of clusters.

[0084] In some embodiments, the software module can be configured to apply the segmentation machine-learning model in can be further can be configured to perform operations can comprise determining a cell boundary of each of the one or more cells of the biological sample. In some embodiments, the software module can be configured to apply the segmentation machine- learning model in can be further configured to perform operations that can comprise determining a cell boundary based at least in part on the combined data can comprise the assessed spatial flows and the cell probability of each cluster of the plurality of clusters. In some embodiments, the software module can be configured to apply the segmentation machine-learning model, which can be further configured to perform operations that can comprise generating the cell segmentation mask based at least in part on the cell boundary of the one or more cells of the biological sample. In some embodiments, the software module can be configured to apply the segmentation machine-learning model to perform operations that can comprise generating one or more cell labels based at least in part on the data. In some embodiments, the software moduleWSGR Docket No.68840-711.601 can be configured to apply the segmentation machine-learning model to perform operations that can comprise generating the cell segmentation mask based at least in part on the generated one or more cell labels.

[0085] In some embodiments, the system can further comprise a software module configured to apply the analysis machine-learning algorithm to segment cells in the biological sample.

[0086] In some embodiments, the analysis machine-learning algorithm can comprise normalizing and aggregating, by the computer processor, the pixel level expression of the plurality of clusters of at least one subset of the plurality of markers. In some embodiments, the analysis machine-learning algorithm can comprise analyzing, by the computer processor, the combined pixel level expression of each of the plurality of clusters. In some embodiments, the analysis machine-learning algorithm can comprise assessing spatial flows and cell probability of each of the plurality of clusters. In some embodiments, the assessment can comprise applying the analysis machine-learning algorithm based on the combined pixel level expression. In some embodiments, the analysis machine-learning algorithm can comprise determining cell boundary of each of the plurality of cells in the biological sample. In some embodiments, the cell boundary can be determined by the computer processor and based on the combined spatial flows and cell probability of each of the plurality of clusters.

[0087] In some embodiments, the software module can be configured to apply the analysis machine-learning algorithm. In some embodiments, the software module can be configured to remove cells with small area or low confidence in segmentation prediction. In some embodiments, the software module can be configured to apply the analysis machine-learning algorithm to merge two different cells. In some embodiments, the cells can be merged based on the area of intersection region.

[0088] In some embodiments, the software module can be configured to apply the analysis machine-learning algorithm to split a cell based on the area of intersection region. In some embodiments, pixel level expression data can comprise protein expression data, RNA expression data, or a combination thereof. In some embodiments, determining a degree of correlation between the pixel level expression and the spatial distribution of the plurality of markers can further comprise analyzing transcriptomic data, proteomic data, or both. In some embodiments, the transcriptomic data, proteomic data, or both can comprise the human protein atlas database. In some embodiments, the transcriptomic data, proteomic data, or both can comprise a gene expression database, an epigenetic database, a biomarker database, or any combination thereof.WSGR Docket No.68840-711.601

[0089] In some embodiments, the system can further comprise a software module configured to optimize cell segmentation. In some embodiments, the cell segmentation can be optimized by applying a segmentation machine-learning model. In some embodiments, the segmentation machine-learning model can utilize a Cellpose-style model, a Mesmer-style model, a StarDist model, or a deep watershed-based model.

[0090] In some embodiments, the plurality of markers can comprise at least about 5, at least about 10, at least about 20, at least about 30, at least about 50, at least about 100, at least about 200 or at least about 300 different markers.. In some embodiments, the plurality of markers can comprise at least about 2, at least about 3, at least about 4, at least about 5, at least about 6, at least about 7, at least about 8, at least about 9, at least about 10, at least about 11, at least about 12, at least about 13, at least about 14, at least about 15, at least about 16, at least about 17, at least about 18, at least about 19, at least about 20, at least about 21, at least about 22, at least about 23, at least about 24, at least about 25, at least about 26, at least about 27, at least about 28, at least about 29, at least about 30, at least about 31, at least about 32, at least about 33, at least about 34, at least about 35, at least about 36, at least about 37, at least about 38, at least about 39, at least about 40, at least about 41, at least about 42, at least about 43, at least about 44, at least about 45, at least about 46, at least about 47, at least about 48, at least about 49, at least about 50, at least about 51, at least about 52, at least about 53, at least about 54, at least about 55, at least about 56, at least about 57, at least about 58, at least about 59, at least about 60, at least about 61, at least about 62, at least about 63, at least about 64, at least about 65, at least about 66, at least about 67, at least about 68, at least about 69, at least about 70, at least about 71, at least about 72, at least about 73, at least about 74, at least about 75, at least about 76, at least about 77, at least about 78, at least about 79, at least about 80, at least about 81, at least about 82, at least about 83, at least about 84, at least about 85, at least about 86, at least about 87, at least about 88, at least about 89, at least about 90, at least about 91, at least about 92, at least about 93, at least about 94, at least about 95, at least about 96, at least about 97, at least about 98, at least about 99, or at least about 100 different markers. In some embodiments, the plurality of markers can comprise at least about 100, at least about 110, at least about 120, at least about 130, at least about 140, at least about 150, at least about 160, at least about 170, at least about 180, at least about 190, at least about 200, at least about 210, at least about 220, at least about 230, at least about 240, at least about 250, at least about 260, at least about 270, at least about 280, at least about 290, at least about 300, at least about 310, at least about 320, at least about 330, at least about 340, at least about 350, at least about 360, at least about 370, at least about 380, at least about 390, at least about 400, at least about 410, at least about 420, atWSGR Docket No.68840-711.601 least about 430, at least about 440, at least about 450, at least about 460, at least about 470, at least about 480, at least about 490, or at least about 500 different markers, including increments therein.

[0091] In some embodiments, the subcellular component can comprise a nucleus, a cell membrane, a cytoplasm, a soma, a protrusion, a mitochondria, a Golgi apparatus, a ribosome, an endoplasmic reticulum, a lysosome, a nucleolus, a nuclear speckles, a nucleoplasm, a nuclear periphery, a Cajal bodies, or a promyelocytic leukemia nuclear (PML) body.

[0092] In some embodiments, the tissue type can comprise brain, prostate, lung, colon, breast, ovary, uterus, esophagus, stomach, liver, kidney, skin, heart, muscle, or bone marrow.. In some embodiments, the tissue type can comprise epithelial tissue, connective tissue, muscle tissue, nervous tissue, blood tissue, bone tissue, adipose tissue, cartilage tissue, cardiac muscle tissue, skeletal muscle tissue, smooth muscle tissue, lymphatic tissue, vascular tissue, or any combination thereof.

[0093] In some embodiments, the plurality of cells can comprise one or more cell types selected from a neuron, an astrocyte, a microglial, an immune cell, an endothelial cell, an epithelial cell, a hepatocyte, a keratinocyte, a myoepithelial cell, or a muscle cell. In some embodiments, the one or more cell types can comprise neuronal cells, muscle cells, adipocytes, hepatocytes, leukocytes, erythrocytes, osteocytes, fibroblasts, keratinocytes, endothelial cells, lymphocytes, phagocytes, cardiomyocytes, nephrons, alveolar cells, enterocytes, astrocytes, oocytes, spermatocytes, zygote, stem cells, induced pluripotent stem cells, epithelial cells, pancreatic cells, retinal cells, hair cells, or any combination thereof.

[0094] In some embodiments the application can further comprise generating, by the computer processor, one or more raw cell probability images. In some embodiments the application can further comprise assigning a value of zero to each existing cell label of the plurality of markers. In some embodiments the application can further comprise identifying one or more combined cell clusters.

[0095] In some embodiments, the operations can further comprise applying a threshold value for assessing spatial flows and cell probability of the one or more combined cell clusters. In some embodiments, the threshold value can be (1x10-1, max), (1x10-2, max), (1x10-3, max), (1x10-4, max), (1x10-5, max), (1x10-6, max), (1x10-7, max), (1x10-8, max), (1x10-9, max), (1x10-10, max), (1x10-11, max), (1x10-12, max), (1x10-13, max), (1x10-14, max), (1x10-15, max), or less than (1x10-15, max). In some embodiments, the threshold value can be (1x10-5, max).WSGR Docket No.68840-711.601

[0096] In some embodiments the application can further comprise applying a size-range filter using the threshold value. In some embodiments the application can further comprise generating one or more new cell labels for the one or more combined cell clusters. In some embodiments, the one or more new cell labels can comprise a marker. In some embodiments, identifying one or more combined cell clusters can comprise setting a threshold value of (0, max) or (-2, max) for assessing spatial flows and cellular probability of the one or more combined clusters.

[0097] In some embodiments the application can further comprise applying a minimum size filter to each combined cluster of the one or more combined clusters. In some embodiments the application can further comprise generating the one or more new cell labels can further comprise reassigning one or more of the existing cell labels. In some embodiments, reassigning the one or more labels can comprise determining the intersection of area (IoA) value of each combined cluster of the one or more combined clusters. In some embodiments, the IoA can comprise the touch IoA of each combined cluster of the one or more combined clusters. In some embodiments the application can further comprise removing one or more of the labels from each combined cluster of the one or more combined clusters having an IoA threshold value of less than 0.6. In some embodiments, the IoA threshold can be 1.0, 0.9, 0.8, 0.7, 0.6, 0.5, 0.4, 0.3, 0.2, 0.1, or less than 0.1.

[0098] In some embodiments the application can further comprise selecting one or more of the existing labels to reassign by applying a nearest neighbor split algorithm. In some embodiments, the nearest neighbor split algorithm can be performed against the existing cell labels and the new cell labels. In some embodiments, the nearest neighbor split algorithm can be performed to generate one or more newly detected cell borders.

[0099] In some embodiments the application can further comprise generating the one or more new cell labels for one or more newly detected cells which can comprise the one or more newly detected cell borders. In some embodiments the application can further comprise generating final cell labels by combining the new cell labels and the existing cell labels. In some embodiments the application can further comprise generating a segmentation mask which can comprise the final cell labels.

[0100] In yet another embodiment, disclosed herein is a computer-implemented method for improving cell segmentation in a biological sample, the method can comprise receiving data by the computer processor. In some embodiments, the data can comprise a plurality of previously identified clustered cellular images. In some embodiments, the method can comprise iteratively applying a machine-learning model to each previously identified cluster of the plurality ofWSGR Docket No.68840-711.601 previously identified clusters. In some embodiments, the method can comprise normalizing and aggregating the pixel level expression of the marker of the previously identified cluster. In some embodiments, the method can comprise analyzing the combined pixel level expression of the previously identified cluster. In some embodiments, the method can comprise assessing spatial flows and cell probability of the previously identified cluster. In some embodiments, the analysis can be based on the combined pixel level expression. In some embodiments, the method can comprise repeating the method steps for each previously identified cluster of the plurality of previously identified clusters. In some embodiments, the method can comprise combining data comprising the assessed spatial flows and cell probability of each previously identified cluster of the plurality of previously identified clusters. In some embodiments, the method can comprise generating a cell segmentation mask based at least in part on the data comprising the assessed spatial flows and cell probability of each previously identified cluster of the plurality of previously identified clusters.

[0101] In yet another embodiment, disclosed herein is a computer-implemented system comprising a computing device comprising at least one processor and instructions executable by the at least one processor to provide an application for improved cell segmentation in a biological sample. In some embodiments, the system can comprise a software module which can be configured to receive input comprising a plurality of previously identified clustered cellular images. In some embodiments, the system can comprise a software module which can be configured to iteratively apply a segmentation machine-learning model to each cluster of the plurality of previously identified clusters. In some embodiments, the system can comprise a software module which can be configured to perform processes comprising normalizing and aggregating the pixel level expression of the marker of the previously identified cluster. In some embodiments, the system can comprise a software module which can be configured to analyze, by the computer processor, the combined pixel level expression of the previously identified cluster. In some embodiments, the system can comprise a software module which can be configured to assess spatial flows and cell probability of the previously identified cluster, based on the combined pixel level expression. In some embodiments, the system can comprise a software module which can be configured to repeat the assessment for each previously identified cluster of the plurality of previously identified clusters. In some embodiments, the system can comprise a software module which can be configured to combine data comprising the assessed spatial flows and cell probability of each previously identified cluster of the plurality of previously identified clusters. In some embodiments, the system can comprise a software module which can be configured to perform operations comprising generating a cellWSGR Docket No.68840-711.601 segmentation mask based at least in part on the data. In some embodiments, the data can comprise the assessed spatial flows and cell probability of each previously identified cluster of the plurality of previously identified clusters.

[0102] In some embodiments, the method can further comprise generating, by the computer processor, one or more raw cell probability images. In some embodiments, the method further can comprise assigning a value of zero to each existing cell label of the plurality of markers. In some embodiments, the method further can comprise identifying one or more combined cell clusters. In some embodiments, the method further can comprise applying a threshold value for assessing spatial flows and cell probability of the one or more combined cell clusters. In some embodiments, the threshold value can be (1x10-5, max). In some embodiments, the threshold value can be 1x10-5, 1x10-6, 1x10-7, 1x10-8, 1x10-9, or 1x10-10, max. In some embodiments, the threshold value can be between -15 and 15, max. In some embodiments, the method further can comprise applying a size-range filter using the threshold value. In some embodiments, the method further can comprise generating one or more new cell labels for the one or more combined cell clusters. In some embodiments, the one or more new cell labels can comprise a marker.

[0103] In some embodiments, the method can further comprise identifying one or more combined cell clusters setting a threshold value for assessing spatial flows and cellular probability of the one or more combined clusters. In some embodiments, the clustering threshold value can be, for example, of (0, max) or (-2, max). In some embodiments, the clustering threshold value can be -15, -14, -13, -12, -11, -10, -9, -8, -7, -6, -5, -4, -3, -2, -1, 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, max. In some embodiments, the method can further comprise applying a minimum size filter to each combined cluster of the one or more combined clusters. In some embodiments, generating the one or more new cell labels can further comprise reassigning one or more of the existing cell labels. In some embodiments, reassigning the one or more labels can comprise determining the intersection of area (IoA) value of each combined cluster of the one or more combined clusters. In some embodiments, the IoA can comprise the touch IoA of each combined cluster of the one or more combined clusters.

[0104] In some embodiments, the method can further comprise removing one or more of the labels from each combined cluster of the one or more combined clusters below an IoA threshold value. In some embodiments, the IoA threshold value can be less than 0.6. In some embodiments, the IoA threshold value can be less than -2.0, -1.5, -1.0, -0.5, 0.0, 0.1, 0.2, 0.3, 0.4, 0.5, 0.6, 0.7, 0.8, 0.9, 1.0, 1.5, 2.0, or more than 2.0. In some embodiments, the method can further comprise selecting one or more of the existing labels to reassign by applying a nearestWSGR Docket No.68840-711.601 neighbor split algorithm. In some embodiments, the nearest neighbor split algorithm can be performed against the existing cell labels and the new cell labels. In some embodiments, the nearest neighbor split algorithm can be performed to generate one or more newly detected cell borders. In some embodiments, the method can further comprise generating the one or more new cell labels for one or more newly detected cells. In some embodiments, the one or more new detected cells can be detected using the one or more newly detected cell borders. In some embodiments, the method can further comprise generating final cell labels by combining the new cell labels and the existing cell labels. In some embodiments, the method can further comprise generating a segmentation mask comprising the final cell labels.

[0105] Computing system

[0106] Referring to FIG.1, a block diagram is shown depicting an exemplary machine that includes a computer system 100 (e.g., a processing or computing system) within which a set of instructions can execute for causing a device to perform or execute any one or more of the aspects and / or methodologies for static code scheduling of the present disclosure. The components in FIG.1 are examples only and do not limit the scope of use or functionality of any hardware, software, embedded logic component, or a combination of two or more such components implementing particular embodiments.

[0107] Computer system 100 may include one or more processors 101, a memory 103, and a storage 108 that communicate with each other, and with other components, via a bus 140. The bus 140 may also link a display 132, one or more input devices 133 (which may, for example, include a keypad, a keyboard, a mouse, a stylus, etc.), one or more output devices 134, one or more storage devices 135, and various tangible storage media 136. All of these elements may interface directly or via one or more interfaces or adaptors to the bus 140. For instance, the various tangible storage media 136 can interface with the bus 140 via storage medium interface 126. Computer system 100 may have any suitable physical form, including but not limited to one or more integrated circuits (ICs), printed circuit boards (PCBs), mobile handheld devices (such as mobile telephones or PDAs), laptop or notebook computers, distributed computer systems, computing grids, or servers.

[0108] Computer system 100 includes one or more processor(s) 101 (e.g., central processing units (CPUs), general purpose graphics processing units (GPGPUs), or quantum processing units (QPUs)) that carry out functions. Processor(s) 101 optionally contains a cache memory unit 102 for temporary local storage of instructions, data, or computer addresses. Processor(s) 101 are configured to assist in execution of computer readable instructions. Computer system 100 mayWSGR Docket No.68840-711.601 provide functionality for the components depicted in FIG.1 as a result of the processor(s) 101 executing non-transitory, processor-executable instructions embodied in one or more tangible computer-readable storage media, such as memory 103, storage 108, storage devices 135, and / or storage medium 136. The computer-readable media may store software that implements particular embodiments, and processor(s) 101 may execute the software. Memory 103 may read the software from one or more other computer-readable media (such as mass storage device(s) 135, 136) or from one or more other sources through a suitable interface, such as network interface 120. The software may cause processor(s) 101 to carry out one or more processes or one or more steps of one or more processes described or illustrated herein. Carrying out such processes or steps may include defining data structures stored in memory 103 and modifying the data structures as directed by the software.

[0109] The memory 103 may include various components (e.g., machine readable media) including, but not limited to, a random access memory component (e.g., RAM 104) (e.g., static RAM (SRAM), dynamic RAM (DRAM), ferroelectric random access memory (FRAM), phase- change random access memory (PRAM), etc.), a read-only memory component (e.g., ROM 105), and any combinations thereof. ROM 105 may act to communicate data and instructions unidirectionally to processor(s) 101, and RAM 104 may act to communicate data and instructions bidirectionally with processor(s) 101. ROM 105 and RAM 104 may include any suitable tangible computer-readable media described below. In one example, a basic input / output system 106 (BIOS), including basic routines that help to transfer information between elements within computer system 100, such as during start-up, may be stored in the memory 103.

[0110] Fixed storage 108 is connected bidirectionally to processor(s) 101, optionally through storage control unit 107. Fixed storage 108 provides additional data storage capacity and may also include any suitable tangible computer-readable media described herein. Storage 108 may be used to store operating system 109, executable(s) 110, data 111, applications 112 (application programs), and the like. Storage 108 can also include an optical disk drive, a solid-state memory device (e.g., flash-based systems), or a combination of any of the above. Information in storage 108 may, in appropriate cases, be incorporated as virtual memory in memory 103.

[0111] In one example, storage device(s) 135 may be removably interfaced with computer system 100 (e.g., via an external port connector (not shown)) via a storage device interface 125. Particularly, storage device(s) 135 and an associated machine-readable medium may provide non-volatile and / or volatile storage of machine-readable instructions, data structures, program modules, and / or other data for the computer system 100. In one example, software may reside,WSGR Docket No.68840-711.601 completely or partially, within a machine-readable medium on storage device(s) 135. In another example, software may reside, completely or partially, within processor(s) 101.

[0112] Bus 140 connects a wide variety of subsystems. Herein, reference to a bus may encompass one or more digital signal lines serving a common function, where appropriate. Bus 140 may be any of several types of bus structures including, but not limited to, a memory bus, a memory controller, a peripheral bus, a local bus, and any combinations thereof, using any of a variety of bus architectures. As an example and not by way of limitation, such architectures include an Industry Standard Architecture (ISA) bus, an Enhanced ISA (EISA) bus, a Micro Channel Architecture (MCA) bus, a Video Electronics Standards Association local bus (VLB), a Peripheral Component Interconnect (PCI) bus, a PCI-Express (PCI-X) bus, an Accelerated Graphics Port (AGP) bus, HyperTransport (HTX) bus, serial advanced technology attachment (SATA) bus, and any combinations thereof.

[0113] Computer system 100 may also include an input device 133. In one example, a user of computer system 100 may enter commands and / or other information into computer system 100 via input device(s) 133. Examples of an input device(s) 133 include, but are not limited to, an alpha-numeric input device (e.g., a keyboard), a pointing device (e.g., a mouse or touchpad), a touchpad, a touch screen, a multi-touch screen, a joystick, a stylus, a gamepad, an audio input device (e.g., a microphone, a voice response system, etc.), an optical scanner, a video or still image capture device (e.g., a camera), and any combinations thereof. In some embodiments, the input device is a Kinect, Leap Motion, or the like. Input device(s) 133 may be interfaced to bus 140 via any of a variety of input interfaces 123 (e.g., input interface 123) including, but not limited to, serial, parallel, game port, USB, FIREWIRE, THUNDERBOLT, or any combination of the above.

[0114] In particular embodiments, when computer system 100 is connected to network 130, computer system 100 may communicate with other devices, specifically mobile devices and enterprise systems, distributed computing systems, cloud storage systems, cloud computing systems, and the like, connected to network 130. Communications to and from computer system 100 may be sent through network interface 120. For example, network interface 120 may receive incoming communications (such as requests or responses from other devices) in the form of one or more packets (such as Internet Protocol (IP) packets) from network 130, and computer system 100 may store the incoming communications in memory 103 for processing. Computer system 100 may similarly store outgoing communications (such as requests or responses to other devices) in the form of one or more packets in memory 103 and communicated to network 130WSGR Docket No.68840-711.601 from network interface 120. Processor(s) 101 may access these communication packets stored in memory 103 for processing.

[0115] Examples of the network interface 120 include, but are not limited to, a network interface card, a modem, and any combination thereof. Examples of a network 130 or network segment 130 include, but are not limited to, a distributed computing system, a cloud computing system, a wide area network (WAN) (e.g., the Internet, an enterprise network), a local area network (LAN) (e.g., a network associated with an office, a building, a campus or other relatively small geographic space), a telephone network, a direct connection between two computing devices, a peer-to-peer network, and any combinations thereof. A network, such as network 130, may employ a wired and / or a wireless mode of communication. In general, any network topology may be used.

[0116] Information and data can be displayed through a display 132. Examples of a display 132 include, but are not limited to, a cathode ray tube (CRT), a liquid crystal display (LCD), a thin film transistor liquid crystal display (TFT-LCD), an organic liquid crystal display (OLED) such as a passive-matrix OLED (PMOLED) or active-matrix OLED (AMOLED) display, a plasma display, and any combinations thereof. The display 132 can interface to the processor(s) 101, memory 103, and fixed storage 108, as well as other devices, such as input device(s) 133, via the bus 140. The display 132 is linked to the bus 140 via a video interface 122, and transport of data between the display 132 and the bus 140 can be controlled via the graphics control 121. In some embodiments, the display is a video projector. In some embodiments, the display is a head- mounted display (HMD) such as a VR headset. In further embodiments, suitable VR headsets include, by way of non-limiting examples, HTC Vive, Oculus Rift, Samsung Gear VR, Microsoft HoloLens, Razer OSVR, FOVE VR, Zeiss VR One, Avegant Glyph, Freefly VR headset, and the like. In still further embodiments, the display is a combination of devices such as those disclosed herein.

[0117] In addition to a display 132, computer system 100 may include one or more other peripheral output devices 134 including, but not limited to, an audio speaker, a printer, a storage device, and any combinations thereof. Such peripheral output devices may be connected to the bus 140 via an output interface 124. Examples of an output interface 124 include, but are not limited to, a serial port, a parallel connection, a USB port, a FIREWIRE port, a THUNDERBOLT port, and any combinations thereof.

[0118] In addition or as an alternative, computer system 100 may provide functionality as a result of logic hardwired or otherwise embodied in a circuit, which may operate in place of orWSGR Docket No.68840-711.601 together with software to execute one or more processes or one or more steps of one or more processes described or illustrated herein. Reference to software in this disclosure may encompass logic, and reference to logic may encompass software. Moreover, reference to a computer-readable medium may encompass a circuit (such as an IC) storing software for execution, a circuit embodying logic for execution, or both, where appropriate. The present disclosure encompasses any suitable combination of hardware, software, or both.

[0119] Those of skill in the art will appreciate that the various illustrative logical blocks, modules, circuits, and model steps described in connection with the embodiments disclosed herein may be implemented as electronic hardware, computer software, or combinations of both. To clearly illustrate this interchangeability of hardware and software, various illustrative components, blocks, modules, circuits, and steps have been described above generally in terms of their functionality.

[0120] The various illustrative logical blocks, modules, and circuits described in connection with the embodiments disclosed herein may be implemented or performed with a general purpose processor, a digital signal processor (DSP), an application specific integrated circuit (ASIC), a field programmable gate array (FPGA) or other programmable logic device, discrete gate or transistor logic, discrete hardware components, or any combination thereof designed to perform the functions described herein. A general purpose processor may be a microprocessor, but in the alternative, the processor may be any conventional processor, controller, microcontroller, or state machine. A processor may also be implemented as a combination of computing devices, e.g., a combination of a DSP and a microprocessor, a plurality of microprocessors, one or more microprocessors in conjunction with a DSP core, or any other such configuration.

[0121] The steps of a method or model described in connection with the embodiments disclosed herein may be embodied directly in hardware, in a software module executed by one or more processor(s), or in a combination of the two. A software module may reside in RAM memory, flash memory, ROM memory, EPROM memory, EEPROM memory, registers, hard disk, a removable disk, a CD-ROM, or any other form of storage medium known in the art. An exemplary storage medium is coupled to the processor such the processor can read information from, and write information to, the storage medium. In the alternative, the storage medium may be integral to the processor. The processor and the storage medium may reside in an ASIC. The ASIC may reside in a user terminal. In the alternative, the processor and the storage medium may reside as discrete components in a user terminal.WSGR Docket No.68840-711.601

[0122] In accordance with the description herein, suitable computing devices include, by way of non-limiting examples, cloud computing platforms, distributed computing platforms, server clusters, server computers, desktop computers, laptop computers, notebook computers, sub- notebook computers, netbook computers, and netpad computers.

[0123] In some embodiments, the computing device includes an operating system configured to perform executable instructions. The operating system is, for example, software, including programs and data, which manages the device’s hardware and provides services for execution of applications. Those of skill in the art will recognize that suitable server operating systems include, by way of non-limiting examples, FreeBSD, OpenBSD, NetBSD®, Linux, Apple® Mac OS X Server®, Oracle® Solaris®, Windows Server®, and Novell® NetWare®. Those of skill in the art will recognize that suitable personal computer operating systems include, by way of non-limiting examples, Microsoft® Windows®, Apple® Mac OS X®, UNIX®, and UNIX- like operating systems such as GNU / Linux®. In some embodiments, the operating system is provided by cloud computing. Those of skill in the art will also recognize that suitable mobile smartphone operating systems include, by way of non-limiting examples, Nokia® Symbian® OS, Apple® iOS®, Research In Motion® BlackBerry OS®, Google® Android®, Microsoft® Windows Phone® OS, Microsoft® Windows Mobile® OS, Linux®, and Palm® WebOS®. Non-transitory computer readable storage medium

[0124] In some embodiments, the platforms, systems, media, and methods disclosed herein include one or more non-transitory computer readable storage media encoded with a program including instructions executable by the operating system of an optionally networked computing device. In further embodiments, a computer readable storage medium is a tangible component of a computing device. In still further embodiments, a computer readable storage medium is optionally removable from a computing device. In some embodiments, a computer readable storage medium includes, by way of non-limiting examples, CD-ROMs, DVDs, flash memory devices, solid state memory, magnetic disk drives, magnetic tape drives, optical disk drives, distributed computing systems including cloud computing systems and services, and the like. In some cases, the program and instructions are permanently, substantially permanently, semi- permanently, or non-transitorily encoded on the media. Computer program

[0125] In some embodiments, the platforms, systems, media, and methods disclosed herein include at least one computer program, or use of the same. A computer program includes a sequence of instructions, executable by one or more processor(s) of the computing device’sWSGR Docket No.68840-711.601 CPU, written to perform a specified task. Computer readable instructions may be implemented as program modules, such as functions, objects, Application Programming Interfaces (APIs), computing data structures, and the like, which perform particular tasks or implement particular abstract data types. In light of the disclosure provided herein, those of skill in the art will recognize that a computer program may be written in various versions of various languages.

[0126] The functionality of the computer readable instructions may be combined or distributed as desired in various environments. In some embodiments, a computer program comprises one sequence of instructions. In some embodiments, a computer program comprises a plurality of sequences of instructions. In some embodiments, a computer program is provided from one location. In other embodiments, a computer program is provided from a plurality of locations. In various embodiments, a computer program includes one or more software modules. In various embodiments, a computer program includes, in part or in whole, one or more web applications, one or more mobile applications, one or more standalone applications, one or more web browser plug-ins, extensions, add-ins, or add-ons, or combinations thereof. Web application

[0127] In some embodiments, a computer program includes a web application. In light of the disclosure provided herein, those of skill in the art will recognize that a web application, in various embodiments, utilizes one or more software frameworks and one or more database systems. In some embodiments, a web application is created upon a software framework such as Microsoft®.NET or Ruby on Rails (RoR). In some embodiments, a web application utilizes one or more database systems including, by way of non-limiting examples, relational, non-relational, object oriented, associative, XML, and document oriented database systems. In further embodiments, suitable relational database systems include, by way of non-limiting examples, Microsoft®SQL Server, mySQL™, and Oracle®. Those of skill in the art will also recognize that a web application, in various embodiments, is written in one or more versions of one or more languages. A web application may be written in one or more markup languages, presentation definition languages, client-side scripting languages, server-side coding languages, database query languages, or combinations thereof. In some embodiments, a web application is written to some extent in a markup language such as Hypertext Markup Language (HTML), Extensible Hypertext Markup Language (XHTML), or eXtensible Markup Language (XML). In some embodiments, a web application is written to some extent in a presentation definition language such as Cascading Style Sheets (CSS). In some embodiments, a web application is written to some extent in a client-side scripting language such as Asynchronous JavaScript and XML (AJAX), Flash®ActionScript, JavaScript, or Silverlight®. In some embodiments, a webWSGR Docket No.68840-711.601 application is written to some extent in a server-side coding language such as Active Server Pages (ASP), ColdFusion®, Perl, Java™, JavaServer Pages (JSP), Hypertext Preprocessor (PHP), Python™, Ruby, Tcl, Smalltalk, WebDNA®, or Groovy. In some embodiments, a web application is written to some extent in a database query language such as Structured Query Language (SQL). In some embodiments, a web application integrates enterprise server products such as IBM®Lotus Domino®. In some embodiments, a web application includes a media player element. In various further embodiments, a media player element utilizes one or more of many suitable multimedia technologies including, by way of non-limiting examples, Adobe®Flash®, HTML 5, Apple®QuickTime®, Microsoft®Silverlight®, Java™, and Unity®. Standalone application

[0128] In some embodiments, a computer program includes a standalone application, which is a program that is run as an independent computer process, not an add-on to an existing process, e.g., not a plug-in. Those of skill in the art will recognize that standalone applications are often compiled. A compiler is a computer program(s) that transforms source code written in a programming language into binary object code such as assembly language or machine code. Suitable compiled programming languages include, by way of non-limiting examples, C, C++, Objective-C, COBOL, Delphi, Eiffel, Java™, Lisp, Python™, Visual Basic, and VB .NET, or combinations thereof. Compilation is often performed, at least in part, to create an executable program. In some embodiments, a computer program includes one or more executable complied applications. Software modules

[0129] In some embodiments, the platforms, systems, media, and methods disclosed herein include software, server, and / or database modules, or use of the same. In view of the disclosure provided herein, software modules are created by techniques known to those of skill in the art using machines, software, and languages known to the art. The software modules disclosed herein are implemented in a multitude of ways. In various embodiments, a software module comprises a file, a section of code, a programming object, a programming structure, a distributed computing resource, a cloud computing resource, or combinations thereof. In further various embodiments, a software module comprises a plurality of files, a plurality of sections of code, a plurality of programming objects, a plurality of programming structures, a plurality of distributed computing resources, a plurality of cloud computing resources, or combinations thereof. In various embodiments, the one or more software modules comprise, by way of non- limiting examples, a web application, a mobile application, a standalone application, and aWSGR Docket No.68840-711.601 distributed or cloud computing application. In some embodiments, software modules are in one computer program or application. In other embodiments, software modules are in more than one computer program or application. In some embodiments, software modules are hosted on one machine. In other embodiments, software modules are hosted on more than one machine. In further embodiments, software modules are hosted on a distributed computing platform such as a cloud computing platform. In some embodiments, software modules are hosted on one or more machines in one location. In other embodiments, software modules are hosted on one or more machines in more than one location. Databases

[0130] In some embodiments, the platforms, systems, media, and methods disclosed herein include one or more databases, or use of the same. In view of the disclosure provided herein, those of skill in the art will recognize that many databases are suitable for storage and retrieval of, for example, image, cell, mask, segmentation, model, user, and project information. In various embodiments, suitable databases include, by way of non-limiting examples, relational databases, non-relational databases, object oriented databases, object databases, entity- relationship model databases, associative databases, XML databases, document oriented databases, and graph databases. Further non-limiting examples include SQL, PostgreSQL, MySQL, Oracle, DB2, Sybase, and MongoDB. In some embodiments, a database is Internet- based. In further embodiments, a database is web-based. In still further embodiments, a database is cloud computing-based. In a particular embodiment, a database is a distributed database. In other embodiments, a database is based on one or more local computer storage devices. Processing

[0131] Referring to FIG.2, in some embodiments, an exemplary methodology for high-plex image-based cell segmentation 200 can begin with a normalization step 205 to normalize intensity across color channels. In some embodiments, the normalization step 205 can be instrument-specific. As depicted in the non-limiting example of FIG.2, the methodology can next include conducting a pixel-level correlation study to identify different spatial groups of markers 210. In some embodiments, the spatial groups of markers can be, for example, protein markers. In some embodiments, the pixel-level correlation study 210 can be conducted once per tissue type, and can be applied to new studies of the same tissue type. Further, as depicted in FIG.2 for example, the data can be further normalized on a per target basis based on, for example, bulk expression 215. The normalization on a per target basis 215 can be based on, for example tissue type or prior knowledge. Next, in this example as depicted in FIG.2, theWSGR Docket No.68840-711.601 combined intensity can be calculated for each segment group 220. One or more appropriate ML models can be next applied, per the example of FIG.2, to segment each group 225 in order to generate gradient flows and / or cell probabilities. Next, in the example of FIG.2, the resultant flows and probabilities can be combined across different groups after further normalization and weighting 230 and the combined flows can be segmented into one set of cell labels 235. The exemplary methodology of FIG.2 can conclude with mask clean-up and finalization given cut- offs for area, flows, and / or cell probabilities 240. Finally, optional post-processing steps 245 are described further herein. Post-Processing

[0132] An exemplary post-processing step is illustrated in FIG.3. A post-processing step can be applied to the integrated cell mask, for example: (1) remove cells with small area and / or low confidence in segmentation prediction; or (2) combine information from additional sets of cell label generated with other methods / platform for further refinement through intersection of union (IoU) analysis as shown in FIG.3, which can merge or split two different sets of cell labels based on the status of intersection region. This post-processing step can be customized for different tissue types as illustrated in FIG.4. For example, in case of tissue type lacking multi- nucleated cells, one could define the integrated cell label generated using the framework as general label (denoted as “C” in schematic of FIG.3) and the nuclei-only cell label of same field-of-view as specific label (denoted as “N” in schematic of FIG.3), and then split or merge the general label based on their IoU area ratios as well as distance to nearest specific label to ensure no more than one nucleus per cell in final cell label. As illustrated in FIG.4, various masks can be flattened and combined for various cell tissue types. For example, masks of cells only, combined flow masks, and IoU merged masks can be combined as shown in FIG.4.

[0133] In another example to capture long cellular protrusions like in the case of astrocytes and neurons, as illustrated in FIGS.5A and 5B, conventional image processing methods, for example watershed or adaptive thresholding can convert the channel positive for those protrusions into object label that marks the spatially disconnected stain-positive regions with different object id. Then, IoU analysis can be performed on the integrated cell label as specific label (denoted as “C” in FIG.3) and the paired protrusion-positive label as general label (denoted as “N” in FIG.3). Additionally, as illustrated in FIG.5A, the general label can be assigned and split with respect to the specific label to generate a cell label with long protrusions attached to the source cell. FIG.5A illustrates an example of masks that can be merged to capture long cellular protrusions, for example combined flow masks, IoU merge masks, and post-IoU masks.WSGR Docket No.68840-711.601

[0134] As illustrated by the exemplary results in FIG.5B, results capturing long cellular protrusions can be generated with a Cellpose-style of cell segmentation models as foundation model for each protein colocalization group. The framework can also use also other machine- learning cell segmentation models if the foundation model for a given group generates continuous predictions at pixel level. before example, Mesmer-style of cell segmentation models can generate a spatial map for inner distance of each pixel to its corresponding cell centroid, and the pixel-wise prediction for intracellular pixel class. The prediction output relating to centroids and intracellular pixels can be further processed using a heat diffusion simulation to generate the corresponding spatial flows and cell probability, thus allowing further output combination across groups using same method as described. Validation

[0135] Validation experiments were performed to test the performance of the described framework, the results of which are illustrated in FIG.6 and FIG.7. The validation was performed using CosMx™ Spatial Molecular Imager proteomics and multi-omics datasets. These datasets encompassed 64 or more distinct protein targets and covered a diverse array of tissue types, ranging from the brain and breast to the kidney, skin, and esophagus. Leveraging the rich information embedded in high-plex protein datasets, our novel framework surpassed the capabilities of our previous pipeline, which relied on a limited number of nuclei and membrane markers. As illustrated in FIG.6 by the exemplary validation result data, masks can be validated using different mask layerings, for example a first and second IoU mask merge combining masks for different cell types and masks with membrane markers, for example Nuc- and C902 mask combination, and Astro and C902- mask combination.

[0136] FIG.7 illustrates additional exemplary mask layering data to visualize the differences in normal tissue and malignant tissue for different tissue types using the novel framework. The different tissue types included, for example, cerebrum tissue, skin tissue, kidney tissue, liver tissue, stomach tissue, and esophagus tissue. EXAMPLES

[0137] The following illustrative examples are representative of embodiments of the software applications, systems, and methods described herein and are not meant to be limiting in any way. Data is provided demonstrating rigorous validation of the performance of the described framework using, for example, CosMx™ Spatial Molecular Imager proteomics datasets. These non-limiting exemplary datasets can encompass 64~120 distinct protein targets, and can cover aWSGR Docket No.68840-711.601 diverse array of tissue types, including the brain, breast, kidney, colon, and esophagus for example. Example 1 – Method of robust cell segmentation

[0138] As illustrated in FIG.2, the method of cell segmentation 200 comprised two phrases: analysis and application. First, intensity was normalized across color channels in each instrument as needed 205. In the analysis phase, pixel-level high-plex protein expression profiles representative of the target tissue type were extracted by random sampling of the space. The spot density heatmap from spatial transcriptomics data could also be used as one of input channel. Correlation analysis was performed on the extracted profiles and clusters of spatially colocalized proteins that serve as landmarks for specific cell types and subcellular components were identified 210. The correlations were further normalized on a per target basis based on bulk expression 215. In the application phase, this grouping of the clusters were applied to new datasets of the same tissue type and combined intensity was calculated 220. By normalizing and aggregating signal within each colocalization group, inputs were provided to pre-trained machine-learning models (e.g., Cellpose) and group-specific cell segmentation outputs (spatial flows and cell probability) were obtained 225. Lastly, the outputs from all groups were integrated 230 and the integrated spatial flows 235 were transformed into the integrated cell mask 240. The cell segmentation model was trained, in one example, from a base model specialized on star-like cells as well as data with balanced cell type compositions. Post- processing was performed after mask generation and finalization 245. Example 2 – Post processing of cell segmentation

[0139] Cell masks can be generated by the method shown in FIG.11, for example. Cell masks can be integrated by using a unified pipeline for 2-step mask generation as illustrated in FIG. 11, for example. An optional post-processing step could be applied to the 2-step generated cell mask to (1) remove cells with small area and / or low confidence in segmentation prediction; and (2) combine information from additional sets of cell label generated with other methods / platform for further refinement through intersection of union (IoU) analysis, which merges or split 2 different sets of cell labels based on the status of intersection region. This post- processing step could be customized for different tissue type. For example, as illustrated in FIG. 11, a nuclear label cell mask and a soma label cell mask can be combined into an integrated cell mask after two-step mask generation has been performed. As illustrated in FIG.3, in case of tissue type lacking multi-nucleated cells, the integrated cell label generated was defined using the framework as general label (denoted as “C” in FIG.3) and the nuclei-only cell label of sameWSGR Docket No.68840-711.601 field-of-view as specific label (denoted as “N” in FIG.3). Subsequently, the general label based on their IoU area ratios as well as distance to nearest specific label was merged or split to ensure no more than one nucleus per cell in final cell label, the exemplary results of which are illustrated in FIG.4.

[0140] As illustrated in FIGS.5A and 5B, to segment cells with long cellular protrusions, such as astrocytes and neurons, conventional image processing methods, for example watershed or adaptive thresholding, were used first to convert the channel positive for those protrusions into object label that marks the spatially disconnected stain-positive regions with different object id. The IoU analysis was then performed on the integrated cell label as specific label (denoted as “C” in FIG.3) and the paired protrusion-positive label as general label (denoted as “N” in FIG. 3). The general label was assigned and split with respect to the specific label to get final cell label with long protrusions attached to the source cell as shown in FIG.5A. The exemplary results of this method are illustrated in FIG.5B. Example 3 – Validating the Cell Segmentation System

[0141] Performance of this cell segmentation system was validated using CosMx™ Spatial Molecular Imager proteomics and multi-mics datasets as illustrated in FIGS.6 and 7. These datasets encompassed 64 or more distinct protein targets and covered a diverse array of tissue types, ranging from the brain and breast to the kidney, skin, and esophagus. As illustrated in FIG.6, this system excelled in recovering cells with faint staining, accurately capturing intricate cell morphologies. Exemplary comparative results of this validation process illustrated as a comparison between adjacent healthy tissue sample results and malignant tissue sample results for various tissue types including cerebrum, skin, kidney, liver, stomach, and esophagus tissues are shown in FIG.7. In FIGs.6 and 7, for example, the white area represents a non-cell region in either mask, green areas are cell regions in both masks, magenta areas are cell regions in the combined mask but non-cell regions in the original mask, indicating regions with positive but weak signal in non-nuclei groups, and brown areas indicate a non-cell region in the combined cell mask, but a cell region in the original mask, indicating lack of nuclei, or cell dilation and low flow magnitude for the region of the positive channel only. Example 4 – Flows to Mask Conversion

[0142] Improved cell segmentation outcomes can be achieved by using lower thresholds and improved algorithms to achieve improved flows-to-mask conversions as illustrated in, for example FIGS.8A-8C. In FIG.8A, areas of 3-channel input with circled areas of interest are compared to predicted dP flows of the data and corresponding predicted cell labels. LowerWSGR Docket No.68840-711.601 default flow thresholds can be used to detect smaller cells or depict existing cells as larger, closer to actual size, by capturing an increased membrane area of the cells. This can be achieved through improved algorithms for flows-to-mask conversion as disclosed herein. FIG.8B depicts a magnified area 1 of FIG.8A, and FIG.8C depicts a magnified area 2 of FIG.8A.

[0143] As depicted in, for example FIGS.9A and 9B, cell labels are not properly recovered at any of the exemplary various flow rate thresholds and cell probability thresholds. In FIG.9A, the raw cell probability data and raw dP flows data were combined as previously described with different cell flows thresholds, including flows at 0, flows greater than or equal to 0.2, flows greater than or equal to 0.4, and flows greater than or equal to 0.6 to yield different imaging results to target various cell types and characteristics. The higher flows threshold can be used to differentiate foreground and background cells and remove error by eliminating weaker flows of background cells. However, labeling of cells remains incomplete at all flow levels as illustrated by the missing cellular labeling of FIG.9A. As seen in FIG.9B, for example, cell probability thresholds were also varied, comparing a cell probability threshold of d = -2 to a cell probability threshold of d = 0, producing different dP and cell probability combined masks and different mask results at flows at 0, flows greater than or equal to 0.2, flows greater than or equal to 0.4, and flows greater than or equal to 0.6 imaging different amounts of cell types and characteristics but with incomplete cell labeling.

[0144] As depicted in, for example FIG.10, high-concave cells can present difficulty in flow tracing, and can result in missing cells during flows-to-mask conversion, meaning recovery of the cell labels is incomplete. FIG.10 depicts examples of masks at various flow cutoffs and cell probability cutoffs to address high-concave cells, at cell probability thresholds compared to flows thresholds to illustrate incomplete cell labeling. In FIG.10, for example, cell probability was set to be above a cutoff threshold value at various flow cutoffs of 0, 0.2, 0.4, 0.6, and 0.8, the cell probability cutoff was -2 compared to the cell probability threshold value at flows cutoffs of 0, 0.2, 0.4, 0.6, and 0.8, and the cell probability cutoff was 0, or at the cell threshold at various flow cutoffs of 0, 0.2, 0.4, 0.6, and 0.8. The results shown in FIG.10, for example, indicated that at some cell probability and flow values high-concave cells were missing, as cell label recovery is incomplete using unmodified mask conversion algorithms. Example 5 – Two-Step Mask Generation

[0145] As shown in, for example FIG.11, masks can be generated from flows in two steps. The exemplary method of FIG.11 can comprise, as a first step, generating default cell labels with gradient tracking 1101 according to the method previously discussed herein, wherein, forWSGR Docket No.68840-711.601 example, a cell probability threshold of 0 and a flow threshold of 0.4 can be used to assure cells are not missing in error, as previously described in Example 5. Next, a binary mask 1102 can be created as shown in step 2 of FIG.11, where the cell labels can be revised to set existing cell labels to 0. In step 3 as shown in FIG.11, the binary mask 1102 can be applied to raw cell probability data 1103 to show regions not present in the default cell labels with gradient tracking. In step 4 as shown in exemplary FIG.11, direct thresholding 1104 can be applied to the modified probability data from step 3 using various cell probability thresholds, for example a maximum cell probability threshold of 1x10-5can be used while a value such as 0 or -2 can be used as a cell probability threshold for identifying the foreground region of old and new cells. As seen in 4.1 and 4.2 of FIG.11, new cells were highlighted alone 1106 using this method, and new plus old cells 1105 were also highlighted using the different cell probability values. Size ranges can be adjusted as preferred 1107. As shown in FIG.11, step 5 can comprise adding a size filter 1108, in which the area cutoff can be the range of the typical cell size to avoid capturing either parts of cells or groups of cells. The size filter can be converted to cell labels and added to the set of default cell labels. In some cases, a more stringent size range is preferred 1107. As shown in FIG.11, in step 6, spatially-disconnected objects in foreground binary mask image can be converted to object label images 1109 where each object has a size equivalent to or larger than one full cell, and intersection of area (IoA) analysis can be performed on the object labels and the combined seed cell labels from step 5. As shown in FIG.11, step 7 can be performed in which object labels are removed if: there is no matching, that is, for example the IoA is less than a threshold value, for example 0.6, which can be the intersection over total area of touching cells in the combined seed cell label image 1110. The object labels can also be reassigned if they match with one or more cell labels, splitting the object labels based on the nearest distance to the matched cell labels. The cell labels 1111 were generated as shown in step 7 of FIG.11, by an imaging software and were not filtered for touch IoA, but can be filtered for touch IoA to refine cell labels and reduce noise.

[0146] FIG.12 depicts an example of 2-step mask generation performed on neuronal tissue, where cells can be highly irregular and relatively discrete. As depicted in FIG.12, input data from 3 channels can be received, along with predicted raw dP flows data, and a default cell labels mask was applied as well as a 2-step mask to achieve results. As shown in FIG.12, the default flows-to-mask conversion can isolate a majority of cells, with missing cells having poor gradient flow tracking results, often occurring in cells with elongated shapes with centroids sitting outside the borders as highlighted with the circles. As shown in FIG.12, isolation of the foreground was achieved and discrete new cells from default cell labels can be selected as wellWSGR Docket No.68840-711.601 as binary masks for combined cells using the 2-step mask generation process. As shown in FIG. 12, touch IoA combination with stringent cutoffs and nearest-neighbor splitting can be used to capture membrane borders for cells that are, for example, existing in default cell labels mask generated data, and also recovering missing outlier cells.

[0147] As shown in FIG.13, 3-channel input was received and default cell labels were applied. Raw cell probability imaging was processed as shown in FIG.11, and these processing results were combined and new labels applied at (1x10-5, max) threshold values and a size-filter in the 1-cell range. The default labels and new labels were then combined, and Voroni distance borders were applied. The combined object labels at a threshold value of (0, max) and a size filter above one cell were then applied. Finally, cell labels were split to generate close-to-final cell labels as shown in FIG.13 in various tissue samplings. Touch IoA combination can be applied to refine cell labels.

[0148] As shown in FIG.14, giant cells such as giant neurons can be missed in default cell labels due to bad flows. For example, in FIG.14, giant neurons were recovered using the two- step mask generation process previously described herein. For example, cells that were split in default cell labels would remain split, and missing tip regions were attached only when not included in the new labels generated. FIG.14 highlights exemplary data depicting giant neurons that are incomplete using default cell labels, but are then complete using the two-step masking process.

[0149] FIG.15 depicts exemplary data resulting from application of the two-step mask method on deconvolution images of cerebellum regions comprising large Purkinje neurons mixed among dense nuclei generated by default cell segmentation models, to recover weak-flow cells.

[0150] FIG.16 depicts exemplary data resulting from application of the two-step mask method on human neuronal cell deconvolution images generated by default cytoplasm cell segmentation models, to recover weak-flow cells.

[0151] FIG.17 depicts exemplary data resulting from application of the two-step mask method on human neuronal cell deconvolution images obtained by default generic nuclear cell segmentation models, to recover weak-flow cells. Example 6 –Multi-Protein-Based Cell Segmentation Using Two-Step Mask Generation

[0152] In some embodiments, for example, two-step masks generation methods can also be used in multi-protein-based cell segmentation as depicted in FIG.18 and FIG.19.WSGR Docket No.68840-711.601

[0153] FIG.18 depicts an exemplary method of using two-step mask generation in multi- protein-based cell segmentation, for example for a protein k in color c channel in a segmentation group g with N proteins, where the model m uses threshold fl for flows and cp for cell probability. First, as shown in FIG.18, first the intensity across color channels can be normalized 1801, by, for example: ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^^^, ^^= ( ^^ ^^ ^^ ^^ ^^ ^^^^, ^^− ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^^^) × ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^^^;

[0154] A pixel-level correlation study can then be performed 1802 to identify different spatial groups of protein markers as shown in FIG.18, and can be performed, for example, once per tissue type, and be directly applicable to new studies of the same tissue type. The intensity can then be further normalized 1803 on a per target basis based on bulk expression, using for example tissue-type-specific prior knowledge by the equation, for example: ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^^^, ^^^^, ^^= ; ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^^^( ^^ ^^ ^^ ^^ ^^ ^^_ ^^ ^^ ^^ ^^)

[0155] The combined intensity can then be calculated 1804 for each segment group, utilizing, for example:

[0156] Each group can be segmented 1805 using appropriate machine learning models involving gradient flows and cell probabilities as shown in FIG.18, and may use the equation, for example: ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^^^( ^^ ^^ ^^ ^^ ^^ ^^ ^^^^, ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^^^, ^^ℎ ^^ ^^ ^^ℎ ^^ ^^ ^^^^ ^^, ^^ ^^) = ^^ ^^ ^^ℎ ^^ ^^ ^^ ^^ ^^^^, ^^× ^^ −min 1 + 0.5 ∗^^, ^^(^^), ^^ ^^ ^^ ≥ ^^ℎ ^^ ^^ℎ ^^ ^^ ^^ ^^ ^^ ^^ ^^ℎ ^^1 × ^^( ^^ ^^ max(^^)−min ( ^^)^^, ^^^^^^ ^^, ^^ ^^^^^^ ^^ ^^ ^^ ^^ ^^^^, ^^), ^^ℎ ^^ ^^ ^^ ^^( ^^^^, ^^) = { ; 0.05, ^^ ^^ ^^^^, ^^< ^^ℎ ^^ ^^ ^^ℎ ^^ ^^ ^^^^ ^^, ^^ ^^

[0157] The flows and cell probabilities can then be combined 1806 as shown in FIG.18, for example, across different groups after further normalization and weighting using the equation, for example:WSGR Docket No.68840-711.601 ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^^^ ^^ ^^ ^^= max( ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^^^, ^^× ^^ ^^ ^^ ^^ℎ ^^2^^) ^^ ^^ ^^ ^^ ^^ ^^, ^^

[0158] The combined flows can then be segmented into one set of cell labels using either default cell segmentation model gradient flow tracking, or two-step mask generation methods 1807 as depicted in FIG.11 and described herein.

[0159] Then, as depicted in FIG.18, mask cleanup and finalization 1808 can take place based on cutoffs for areas, flows, and cell probabilities. The two-step mask generation process can take advantage of clean-representation of cells predicted by machine-learning cell segmentation models to achieve robustness that cannot be achieved with mere thresholding on raw intensity images. Two-step mask generation can achieve better capture of cell membranes and recover more cells missing from a default flows-to-mask procedure. Two-step mask generation can be applied to multiple new segmentation models and various tissue types without introduction of merging or splitting errors.

[0160] FIG.19 depicts an exemplary two-step mask generation workflow integrated with cell segmentation methods. As depicted in FIG.19, for example, input images 1901 can comprise 1- channel nuclear stain data 1905, 2-channel or 3-channel cytology stain data 1911, or other data. Next, as depicted in FIG.19, cell segmentation models for different sample types can be applied, such as any cell segmentation model 1902, Nuclei segmentation models 1906, Cytology segmentation models 1912, and / or Neuronal segmentation models 1915. Then default cell labels 1903 can be applied and raw cell probability images 1907 can be generated. The raw cell probability image 1907 can be set to zero on existing cells 1908, and a threshold of (0, max) or (-2, max) 1913 can be applied with minimum size filters for labeling of combined cell clusters 1916. As depicted in FIG.19, threshold values of (1x10-5, max) 1909 can be applied and a size- range filter 1910 can be applied for new cell labels to achieve labeling of existing plus new cells 1904 as spatially separated seed objects. The label image for combined cell clusters can be reassigned based on the intersection of area (IoA) 1918 between each cluster to all cells it touches in the seed label image. The reassignment process involves removal of cluster labels whose touch IoA is smaller than a threshold value 1918, for example 0.6, along with nearest neighbor split 1917 of a cell cluster label against all its touched seed cell labels to generate final cell labels 1914 as depicted in FIG.19.

[0161] Various tissues can undergo a unified method, with variation on the type of models for specialized input for various cell and tissue types as shown in FIGS.20A and 20B. For example, FIG.20A depicts an exemplary unified pipeline for various segmentation modules for differentWSGR Docket No.68840-711.601 stains. For example, as shown in FIG.20A, nuclei stain scan data 2001 can be processed by a nuclei segmentation model 2002 and can undergo 2-step mask generation 2003 optionally after processing with the nuclei segmentation model 2002. Subsequently to generation of the 2-step mask generation 2003, nuclei data only can be labeled with nuclear labels 2004. Cytoplasm and / or membrane stain scan data 2006 can undergo cytoplasm segmentation modeling 2007. The cytoplasm segmentation model 2007 can also receive nuclei stain data 2001 for processing. As shown in FIG.20A, the neuronal segmentation model 2009 can receive protrusion staining scan data 2008, and optionally cytoplasm and membrane stain data 2006, and optionally nuclei stain data as well 2001. Neuronal segmentation modeling 2009 can achieve better capturing of cells of highly irregular shapes and protrusions observed in neuronal samples, for example using a third imaging channel. As shown in FIG.20A, the cytology segmentation model 2007 and neuronal segmentation model 2009 can enter a separate pipeline from the nuclei segmentation model 2002, undergoing a separate 2-step mask generation step 2010. Additionally, the next step can be labeling the soma and protrusion data using soma labels 2011. The next step can be combining soma labels 2011 and nuclear labels 2004, specifically using IoA of the nuclear data and using IoU 2005 to combine the labeling data. Lastly, as illustrated in FIG.20A, the combination of soma labels 2010 and nuclear labels 2004 using IoU and IoA 2005 can generate final cell labels 2012 with the unified method.

[0162] FIG.20B depicts, in image processing, an exemplary combination of nuclei stain data, cytoplasm or membrane stain data, or both, and protrusion data into imaging of a single cell or separate cells depending on the IoU and IoA value thresholds. Nuclei for each start cell are labeled as N12013 and N22019 where the cells as imaged are labeled C32015 and C42017, respectively. The combined single cell 2014 can have an IoU above 0.4 and an IoA more than 0.9, for example. The separate cell 2016 and separate nucleus 2018 can have an IoU of less than or equal to 0.4, and an IoA of less than or equal to 0.9, for example as depicted in FIG.20B.

[0163] FIG.21 depicts an exemplary non-limiting workflow process 2100 for cell segmentation and processing of pre-identified cluster data. As illustrated in FIG.21 data can be received from a database in previously identified format 2110. In the exemplary workflow illustrated in FIG. 21, the previously identified data can be, for example, previously identified clusters of markers 2110. The previously identified markers can be arranged in unique spatial expression patterns because they are previously identified as being in a unique spatial expression pattern 2110. The previously identified data can comprise intensity correlations on a per target basis, for each type of pre-identified marker 2115. Exemplary data of these pre-identified clusters of markers were further normalized on a per target basis based on bulk expression 2115. In the application phase,WSGR Docket No.68840-711.601 this pre-identified grouping of the clusters was applied to new datasets of the same tissue type and combined intensity was calculated 2120. By normalizing and aggregating signal within each colocalization group, inputs were provided to pre-trained machine-learning models (e.g., Cellpose) and group-specific cell segmentation outputs (spatial flows and cell probability) were obtained 2125. Lastly, the outputs from all groups were integrated 2130 and the integrated spatial flows 2135 were transformed into the integrated cell mask 2140. The cell segmentation model was trained, in one example, from a base model specialized on star-like cells as well as data with balanced cell type compositions. Post-processing was performed after mask generation and finalization 2145. As shown in FIG.21, for example, analysis steps including application of analysis machine learning algorithms are not used because no unidentified raw data is input in this exemplary workflow 2110.

[0164] While preferred embodiments of the present subject matter have been shown and described herein, it will be obvious to those skilled in the art that such embodiments are provided by way of example only. Numerous variations, changes, and substitutions will now occur to those skilled in the art without departing from the present subject matter. It should be understood that various alternatives to the embodiments of the present subject matter described herein may be employed in practicing the present subject matter.

Claims

WSGR Docket No.68840-711.601 CLAIMS WHAT IS CLAIMED IS:

1. A computer-implemented method for improving cell segmentation in a biological sample, the method comprising: (a) receiving grouping data by the computer processor, the grouping data comprising a plurality of previously identified clusters of markers, where each previously identified cluster comprises a unique spatial expression pattern of the markers; (b) normalizing, by the computer processor, the expression of each marker of each previously identified cluster; (c) generating an aggregate intensity of expression of the markers for each previously identified cluster; (d) iteratively applying a segmentation machine-learning model to each cluster of the previously identified clusters of markers, wherein the machine-learning model is configured to perform operations comprising: (i) assessing spatial flows and cell probability of the cluster, based on the aggregate intensity of expression, and (iii) repeating the assessing for each cluster of the plurality of previously identified clusters; (e) combining, by the computer processor, data comprising the assessed spatial flows and cell probability of each cluster of the plurality of previously identified clusters; and (f) generating a cell segmentation mask based at least in part on the data of (e).

2. The computer-implemented method of claim 1, wherein the segmentation machine- learning model comprises one or more types of segmentation machine-learning models.

3. The computer-implemented method of claim 2, further comprising applying a predetermined type of segmentation machine-learning model to each cluster of a subset of the plurality of clusters, wherein the subset of the plurality of clusters comprise the same identifier.

4. The computer-implemented method of claim 3, further comprising independently associating each identifier with a type of segmentation machine-learning model.

5. The computer-implemented method of claim 1, further comprising, in (c), determining, by the computer processor, a cell boundary of each of the one or more cells of the biological sample.WSGR Docket No.68840-711.601 6. The computer-implemented method of claim 5, wherein the cell boundary is determined based at least in part on the combined data comprising the assessed spatial flows and the cell probability of each cluster of the plurality of clusters.

7. The computer-implemented method of any one of claims 1-6, further comprising generating the cell segmentation mask based at least in part on the cell boundary of the one or more cells of the biological sample.

8. The computer-implemented method of any one of claims 1-7, further comprising generating one or more cell labels based at least in part on the data of (d).

9. The computer-implemented method of claim 8, further comprising generating the cell segmentation mask based at least in part on the generated one or more cell labels.

10. The computer-implemented method of any one of claims 1-9, further comprising removing clusters with small area or low confidence in segmentation prediction.

11. The computer-implemented method of any one of claims 1-10, further comprising merging two or more different clusters based on the area of intersection region.

12. The computer-implemented method of any one of claims 1-11, further comprising splitting one or more clusters based on the area of intersection region.

13. The computer-implemented method of any one of claims 1-12, wherein the intensity of expression data comprises proteomics data or transcriptomics data, or both.

14. The computer-implemented method of claim 13, wherein the proteomics data or transcriptomics data, or both, comprise protein expression data, RNA expression data, or both.

15. The computer-implemented method of any one of claims 1-14, wherein normalizing the expression of each marker further comprises analyzing transcriptomic data, proteomic data, or both.

16. The computer-implemented method of claim 15, wherein the transcriptomic data, proteomic data, or both comprise a human protein atlas database.

17. The computer-implemented method of any one of claims 1-16, wherein the segmentation machine-learning model comprises a cellular segmentation model.

18. The computer-implemented method of claim 17, wherein the cellular segmentation model comprises a Cellpose-style model, a Mesmer-style model, a StarDist model, or a deep watershed-based model.WSGR Docket No.68840-711.601 19. The computer-implemented method of any one of claims 1-18, further comprising repeating the method for each marker of a plurality of markers.

20. The computer-implemented method claim 19, wherein the plurality of markers comprises at least about 5, at least about 10, at least about 20, at least about 30, at least about 50, at least about 100, at least about 200 or at least about 300 different markers.

21. The computer-implemented method of any one of claims 1-20, wherein the marker comprises a subcellular component.

22. The computer-implemented method of claim 21, wherein the subcellular component comprises one or more of a nucleus, a cell membrane, a cytoplasm, a soma, a protrusion, a mitochondria, a Golgi apparatus, a ribosome, an endoplasmic reticulum, a lysosome, a nucleolus, a nuclear speckles, a nucleoplasm, a nuclear periphery, a Cajal bodies, or a promyelocytic leukemia nuclear (PML) body.

23. The computer-implemented method of any one of claims 1-22, wherein the biological sample comprises a biological sample derived from or sampled from tissue comprising a brain, prostate, lung, colon, breast, ovary, uterus, esophagus, stomach, liver, kidney, skin, heart, muscle, or bone marrow.

24. The computer-implemented method of any one of claims 1-23, wherein the plurality of cells comprises one or more cell types selected from a neuron, an astrocyte, a microglial, an immune cell, an endothelial cell, an epithelial cell, a hepatocyte, a keratinocyte, a myoepithelial cell, or a muscle cell.

25. The computer-implemented method of any one of claims 1-24, wherein the segmentation machine-learning model comprises one or more types of segmentation machine-learning models.

26. The computer-implemented method of claim 25, further comprising applying a predetermined type of segmentation machine-learning model to each cluster of a subset of the plurality of clusters, wherein the subset of the plurality of clusters comprise the same identifier.

27. The computer-implemented method of claim 26, further comprising independently associating each identifier with a type of segmentation machine-learning model.

28. The computer implemented method of any one of claims 1-27, further comprising generating, by the computer processor, one or more raw cell probability images.

29. The computer implemented method of claim 28, further comprising assigning a value of zero to each existing cell label of the plurality of markers.WSGR Docket No.68840-711.601 30. The computer implemented method of claim 28 or 29, further comprising identifying one or more combined cell clusters.

31. The computer implemented method of any one of claims 28-30, further comprising applying a threshold value for assessing spatial flows and cell probability of the one or more combined cell clusters.

32. The computer-implemented method of claim 31, wherein the threshold value is (1x10-1, max), (1x10-2, max), (1x10-3, max), (1x10-4, max), (1x10-5, max), (1x10-6, max), (1x10-7, max), (1x10-8, max), (1x10-9, max), (1x10-10, max), (1x10-11, max), (1x10-12, max), (1x10-13, max), (1x10-14, max), (1x10-15, max), or less than (1x10-15, max).

33. The computer-implemented method of claim 32, wherein the threshold value is (1x10-5, max).

34. The computer implemented method of any one of claims 31-33, further comprising applying a size-range filter using the threshold value.

35. The computer implemented method of any one of claims 30-34, further comprising generating one or more new cell labels for the one or more combined cell clusters.

36. The computer implemented of claim 35, wherein the one or more new cell labels comprise a marker.

37. The computer implemented method any one of claims 30-36, wherein identifying one or more combined cell clusters comprises setting a threshold value of (0, max) or (-2, max) for assessing spatial flows and cellular probability of the one or more combined clusters.

38. The computer implemented method of any one of claims 30-37, further comprising applying a minimum size filter to each combined cluster of the one or more combined clusters.

39. The computer implemented method of any one of claims 35-38, wherein generating the one or more new cell labels further comprises reassigning one or more of the existing cell labels.

40. The computer implemented method of claim 39, wherein reassigning the one or more labels comprises determining the intersection of area (IoA) value of each combined cluster of the one or more combined clusters.

41. The computer implemented method of claim 40, wherein the IoA comprises the touch IoA of each combined cluster of the one or more combined clusters.WSGR Docket No.68840-711.601 42. The computer implemented method of claim 40 or 41, further comprising removing one or more of the labels from each combined cluster of the one or more combined clusters having an IoA threshold value of less than 0.

6.

43. The computer implemented method of any one of claims 29-42, further comprising selecting one or more of the existing labels to reassign by applying a nearest neighbor split algorithm.

44. The computer implemented method of claim 43, wherein the nearest neighbor split algorithm is performed against the existing cell labels and the new cell labels.

45. The computer implemented method of claim 43 or 44, wherein the nearest neighbor split algorithm is performed to generate one or more newly detected cell borders.

46. The computer implemented method of claim 45, further comprising generating the one or more new cell labels for one or more newly detected cells comprising the one or more newly detected cell borders.

47. The computer implemented method of claim 46, further comprising generating final cell labels by combining the new cell labels and the existing cell labels.

48. The computer implemented method of claim 47, further comprising generating a segmentation mask comprising the final cell labels.

49. A system comprising one or more computer processors and computer memory coupled thereto, the computer memory comprising a machine executable code that, upon execution by the one or more computer processors, implements the method of any one of the preceding claims.

50. A computer-implemented system comprising a computing device comprising at least one processor and instructions executable by the at least one processor to provide an application for improved cell segmentation in a biological sample, the application comprising: (a) a software module configured to receive grouping data comprising a plurality of previously identified clusters of markers, where each previously identified cluster of the plurality of previously identified clusters comprises a unique spatial expression pattern of the markers; (b) a software module configured to normalize the expression of each marker of each previously identified cluster;WSGR Docket No.68840-711.601 (c) a software module configured to generate an aggregate intensity of expression of the markers for each cluster of the plurality of previously identified clusters. (d) a software module configured to iteratively apply a segmentation machine-learning model to each cluster of the plurality of previously identified clusters of markers, wherein the machine-learning model is configured to perform operations comprising: (i) assessing spatial flows and cell probability of the cluster, based on the aggregate intensity of expression of (c), and (iii) repeating the assessing for each cluster of the plurality of previously identified clusters; (e) a software module configured to combine data comprising the assessed spatial flows and cell probability of each cluster of the plurality of previously identified clusters; and (f) a software module configured to perform operations comprising generating a cell segmentation mask based at least in part on the data of (e).

51. The computer-implemented system of claim 50, wherein the segmentation machine- learning model comprises one or more types of segmentation machine-learning models.

52. The computer-implemented system of claim 51, the application further comprising applying a predetermined type of segmentation machine-learning model to each cluster of a subset of the plurality of clusters, wherein the subset of the plurality of clusters comprise the same identifier.

53. The computer-implemented system of claim 52, the application further comprising independently associating each identifier with a type of segmentation machine-learning model.

54. The computer-implemented system of any one of claims 50-53, wherein the operations further comprise, in (c), determining a cell boundary of each of the one or more cells of the biological sample.

55. The computer-implemented system of claim 54, wherein the cell boundary is determined based at least in part on the combined data comprising the assessed spatial flows and the cell probability of each cluster of the plurality of clusters.

56. The computer-implemented system of claim 54 or claim 55, the application further comprising generating the cell segmentation mask based at least in part on the cell boundary of the one or more cells of the biological sample.WSGR Docket No.68840-711.601 57. The computer-implemented system of any one of claims 50-56, the application further comprising generating one or more cell labels based at least in part on the data of (d).

58. The computer-implemented system of any one of claims 50-57, the application further comprising generating the cell segmentation mask based at least in part on the generated one or more cell labels.

59. The computer-implemented system of any one of claims 50-58, the application further comprising removing clusters with small area or low confidence in segmentation prediction.

60. The computer-implemented system of any one of claims 50-59, the application further comprising merging two or more different clusters based on the area of intersection region.

61. The computer-implemented system of any one of claims 50-60, the application further comprising splitting one or more clusters based on the area of intersection region.

62. The computer-implemented system of any one of claims 50-61, wherein the intensity of expression data comprises proteomics data or transcriptomics data, or both.

63. The computer-implemented system of claim 62, wherein the proteomics data or transcriptomics data, or both, comprise protein expression data, RNA expression data, or both.

64. The computer-implemented system of any one of claims 50-63, wherein the operations further comprise normalizing the expression of each marker, wherein normalizing the expression of each marker further comprises analyzing transcriptomic data, proteomic data, or both.

65. The computer-implemented system of claim 64, wherein the transcriptomic data, proteomic data, or both comprise a human protein atlas database.

66. The computer-implemented system of any one of claims 50-65, wherein the segmentation machine-learning model comprises a cellular segmentation model.

67. The computer-implemented system of claim 66, wherein the cellular segmentation model comprises a Cellpose-style model, a Mesmer-style model, a StarDist model, or a deep watershed-based model.

68. The computer-implemented system of any one of claims 50-67, the operations further comprising repeating the system for each marker of a plurality of markers.

69. The computer-implemented system claim 68, wherein the plurality of markers comprises at least about 5, at least about 10, at least about 20, at least about 30, at least about 50, at least about 100, at least about 200 or at least about 300 different markers.WSGR Docket No.68840-711.601 70. The computer-implemented system of any one of claims 50-69, wherein the marker comprises a subcellular component.

71. The computer-implemented system of claim 70, wherein the subcellular component comprises a nucleus, a cell membrane, a cytoplasm, a soma, a protrusion, a mitochondria, a Golgi apparatus, a ribosome, an endoplasmic reticulum, a lysosome, a nucleolus, a nuclear speckles, a nucleoplasm, a nuclear periphery, a Cajal bodies, or a promyelocytic leukemia nuclear (PML) body.

72. The computer-implemented system of any one of claims 50-71, wherein the biological sample comprises a biological sample derived from or sampled from tissue comprising a brain, prostate, lung, colon, breast, ovary, uterus, esophagus, stomach, liver, kidney, skin, heart, muscle, or bone marrow.

73. The computer-implemented system of any one of claims 50-72, wherein the plurality of cells comprises one or more cell types selected from a neuron, an astrocyte, a microglial, an immune cell, an endothelial cell, an epithelial cell, a hepatocyte, a keratinocyte, a myoepithelial cell, or a muscle cell.

74. The computer-implemented system of any one of claims 50-73, the application further comprising generating, by the computer processor, one or more raw cell probability images.

75. The computer-implemented system of any one of claims 50-74, the application further comprising assigning a value of zero to each of one or more existing cell labels of the plurality of markers.

76. The computer-implemented system of any one of claims 50-75, the application further comprising identifying one or more combined cell clusters.

77. The computer-implemented system of claim 76, wherein the operations further comprise applying a threshold value for assessing spatial flows and cell probability of the one or more combined cell clusters.

78. The computer-implemented system of claim 77, wherein the threshold value is (1x10-1, max), (1x10-2, max), (1x10-3, max), (1x10-4, max), (1x10-5, max), (1x10-6, max), (1x10-7, max), (1x10-8, max), (1x10-9, max), (1x10-10, max), (1x10-11, max), (1x10-12, max), (1x10-13, max), (1x10-14, max), (1x10-15, max), or less than (1x10-15, max).

79. The computer-implemented system of claim 78, wherein the threshold value is (1x10-5, max).WSGR Docket No.68840-711.601 80. The computer-implemented system of any one of claims 77-79, wherein the operations further comprise applying a size-range filter using the threshold value.

81. The computer-implemented system of any one of claims 76-80, the application further comprising generating one or more new cell labels for the one or more combined cell clusters.

82. The computer-implemented system of claim 81, wherein the one or more new cell labels comprise a marker.

83. The computer-implemented system of any one of claims 76-82, wherein the operations further comprise identifying one or more combined cell clusters 84. The computer-implemented system of claim 83, wherein the operations further comprise setting a threshold value of (0, max) or (-2, max) for assessing spatial flows and cellular probability of the one or more combined clusters.

85. The computer-implemented system of any one of claims 76-84, wherein the operations further comprise applying a minimum size filter to each combined cluster of the one or more combined clusters.

86. The computer-implemented system of any one of claims 75-85, the application further comprising generating the one or more new cell labels, which further comprises reassigning one or more of the existing cell labels.

87. The computer-implemented system of claim 86, the application further comprising reassigning the one or more labels by determining the intersection of area (IoA) value of each combined cluster of the one or more combined clusters.

88. The computer-implemented system of claim 87, the application further comprising the IoA, which comprises the touch IoA of each combined cluster of the one or more combined clusters.

89. The computer-implemented system of claim 87 or 88, the application further comprising removing one or more of the labels from each combined cluster of the one or more combined clusters having an IoA threshold value of less than 0.

6.

90. The computer-implemented system of any one of claims 75-89, the application further comprising selecting one or more of the existing cell labels to reassign by applying a nearest neighbor split algorithm.WSGR Docket No.68840-711.601 91. The computer-implemented system of claim 90, wherein the application further comprises performing nearest neighbor split algorithms against the existing cell labels and the new cell labels.

92. The computer-implemented system of claim 90 or 91, wherein the application further comprises performing the nearest neighbor split algorithm to generate one or more newly detected cell borders.

93. The computer-implemented system of claim 92, the application further comprising generating the one or more new cell labels for one or more newly detected cells comprising the one or more newly detected cell borders.

94. The computer-implemented system of claim 93, the application further comprising generating final cell labels by combining the new cell labels and the existing cell labels.

95. The computer-implemented system of claim 94, the application further comprising generating a segmentation mask comprising the final cell labels.

96. A computer-implemented method for improving cell segmentation in a biological sample, the method comprising: (a) applying an analysis machine-learning algorithm for clustering, comprising: (i) obtaining a plurality of scanned images, wherein each scanned image of the plurality of scanned images comprises one or more cells, wherein the one or more cells are labeled with a marker, (ii) extracting, by a computer processor, pixel-level expression data of the marker of each scanned image of the plurality of scanned images, (iii) determining, by the computer processor, a degree of correlation between the pixel level expression data and the spatial distribution of the marker of each scanned image of the plurality of scanned images, and (iv) identifying, by the computer processor and based on the degree of correlation determined in (a)(iii), a plurality of clusters of the marker of each scanned image of the plurality of scanned images, wherein each cluster of the plurality of clusters comprises an identifier for a cell type, or a subcellular component, or a tissue characteristic; (b) normalizing, by the computer processor the expression of each marker of each identifier of (a)(iv);WSGR Docket No.68840-711.601 (c) generating an aggregate intensity of expression of the markers for each cluster of the plurality of clusters identified in (a)(iv); (d) iteratively applying a segmentation machine-learning model to each cluster of the plurality of clusters identified in (a)(iv), comprising: (i) assessing spatial flows and cell probability of the cluster, based on the aggregate intensity of expression of (c), and (iii) repeating the assessing for each cluster of the plurality of clusters; (e) combining, by the computer processor, data comprising the assessed spatial flows and cell probability of each cluster of the plurality of clusters; and (f) generating a cell segmentation mask based at least in part on the data of (e).

97. The computer-implemented method of claim 96, wherein the analysis machine-learning algorithm comprises an unsupervised machine-learning algorithm.

98. The computer-implemented method of claim 96 or 97, wherein the analysis machine- learning algorithm comprises an unsupervised clustering algorithm.

99. The computer-implemented method of any one of claims 96-98, wherein the analysis machine-learning algorithm comprises one or more of a Louvian clustering algorithm, a Leiden clustering algorithm, or a K-means clustering algorithm, or any combination thereof.

100. The computer-implemented method of any one of claims 96-99, wherein the segmentation machine-learning model comprises one or more types of segmentation machine- learning models.

101. The computer-implemented method of claim 100, further comprising applying a predetermined type of segmentation machine-learning model to each cluster of a subset of the plurality of clusters, wherein the subset of the plurality of clusters comprise the same identifier.

102. The computer-implemented method of claim 100 or 101, further comprising independently associating each identifier with a type of segmentation machine-learning model.

103. The computer-implemented method of any one of claims 96-102, further comprising, in (c), determining a cell boundary of each of the one or more cells of the biological sample.

104. The computer-implemented method of claim 103, wherein the cell boundary is determined based at least in part on the combined data comprising the assessed spatial flows and the cell probability of each cluster of the plurality of clusters.WSGR Docket No.68840-711.601 105. The computer-implemented method of claim 103 or claim 104, further comprising generating the cell segmentation mask based at least in part on the cell boundary of the one or more cells of the biological sample.

106. The computer-implemented method of any one of claims 96-105, further comprising generating one or more cell labels based at least in part on the data of (d).

107. The computer-implemented method of claim 106, further comprising generating the cell segmentation mask based at least in part on the generated one or more cell labels.

108. The computer-implemented method of any one of claims 96-107, further comprising removing clusters with small area or low confidence in segmentation prediction.

109. The computer-implemented method of any one of claims 96-108, further comprising merging two or more different clusters based on the area of intersection region.

110. The computer-implemented method of any one of claims 96-109, further comprising splitting one or more clusters based on the area of intersection region.

111. The computer-implemented method of any one of claims 96-110, wherein the intensity of expression data comprises proteomics data or transcriptomics data, or both.

112. The computer-implemented method of claim 111, wherein the proteomics data or transcriptomics data, or both, comprises protein expression data, RNA expression data, or both.

113. The computer-implemented method of any one of claims 96-112, further comprising determining, in (a)(iii), a degree of correlation between the pixel level expression and the spatial distribution of the plurality of markers.

114. The computer-implemented method of claim 113, wherein the transcriptomic data, proteomic data, or both comprise a human protein atlas database.

115. The computer-implemented method of any one of claims 96-114, wherein the segmentation machine-learning model comprises a cellular segmentation model.

116. The computer-implemented method of claim 115, wherein the cellular segmentation model comprises a Cellpose-style model, a Mesmer-style model, a StarDist model, or a deep watershed-based model.

117. The computer-implemented method of any one of claims 95-116, further comprising repeating the method for each marker of a plurality of markers.WSGR Docket No.68840-711.601 118. The computer-implemented method of claim 117, wherein the plurality of markers comprises at least about 5, at least about 10, at least about 20, at least about 30, at least about 50, at least about 100, at least about 200 or at least about 300 different markers.

119. The computer-implemented method of any one of claims 96-118, wherein the marker comprises a subcellular component.

120. The computer-implemented method of claim 119, wherein the subcellular component comprises a nucleus, a cell membrane, a cytoplasm, a soma, a protrusion, a mitochondria, a Golgi apparatus, a ribosome, an endoplasmic reticulum, a lysosome, a nucleolus, a nuclear speckles, a nucleoplasm, a nuclear periphery, a Cajal bodies, or a promyelocytic leukemia nuclear (PML) body.

121. The computer-implemented method of any one of claims 96-120, wherein the biological sample comprises a biological sample derived from or sampled from tissue comprising a brain, prostate, lung, colon, breast, ovary, uterus, esophagus, stomach, liver, kidney, skin, heart, muscle, or bone marrow.

122. The computer-implemented method of any one of claims 96-121, wherein the plurality of cells comprises one or more cell types selected from a neuron, an astrocyte, a microglial, an immune cell, an endothelial cell, an epithelial cell, a hepatocyte, a keratinocyte, a myoepithelial cell, or a muscle cell.

123. The computer-implemented method of any one of claims 96-122, further comprising generating, by the computer processor, one or more raw cell probability images.

124. The computer-implemented method of any one of claims 96-123, further comprising assigning a value of zero to each existing cell label of the plurality of markers.

125. The computer-implemented method of any one of claims 96-124, further comprising identifying one or more combined cell clusters.

126. The computer-implemented method of claim 125, further comprising applying a threshold value for assessing spatial flows and cell probability of the one or more combined cell clusters.

127. The computer-implemented method of claim 126, wherein the threshold value is (1x10-1, max), (1x10-2, max), (1x10-3, max), (1x10-4, max), (1x10-5, max), (1x10-6, max), (1x10-7, max), (1x10-8, max), (1x10-9, max), (1x10-10, max), (1x10-11, max), (1x10-12, max), (1x10-13, max), (1x10-14, max), (1x10-15, max), or less than (1x10-15, max).WSGR Docket No.68840-711.601 128. The computer-implemented method of claim 127, wherein the threshold value is (1x10-5, max).

129. The computer-implemented method of claim 128, further comprising applying a size- range filter using the threshold value.

130. The computer-implemented method of any one of claims 125-129, further comprising generating one or more new cell labels for the one or more combined cell clusters.

131. The computer-implemented of claim 130, wherein the one or more new cell labels comprise a marker.

132. The computer-implemented method any one of claims 125-131, wherein identifying one or more combined cell clusters comprises setting a threshold value of (0, max) or (-2, max) for assessing spatial flows and cellular probability of the one or more combined clusters.

133. The computer-implemented method of any one of claims 125-132, further comprising applying a minimum size filter to each combined cluster of the one or more combined clusters.

134. The computer-implemented method of any one of claims 130-133, wherein generating the one or more new cell labels further comprises reassigning one or more of the existing cell labels.

135. The computer-implemented method of claim 134, wherein reassigning the one or more labels comprises determining the intersection of area (IoA) value of each combined cluster of the one or more combined clusters.

136. The computer-implemented method of claim 135, wherein the IoA comprises the touch IoA of each combined cluster of the one or more combined clusters.

137. The computer-implemented method of claim 135 or 136, further comprising removing one or more of the labels from each combined cluster of the one or more combined clusters having an IoA threshold value of less than 0.

6.

138. The computer-implemented method of any one of claims 135-137, further comprising selecting one or more of the existing labels to reassign by applying a nearest neighbor split algorithm.

139. The computer-implemented method of claim 138, wherein the nearest neighbor split algorithm is performed against the existing cell labels and the new cell labels.

140. The computer-implemented method of claim 138 or 139, wherein the nearest neighbor split algorithm is performed to generate one or more newly detected cell borders.WSGR Docket No.68840-711.601 141. The computer-implemented method of claim 140, further comprising generating the one or more new cell labels for one or more newly detected cells comprising the one or more newly detected cell borders.

142. The computer-implemented method of claim 141, further comprising generating final cell labels by combining the new cell labels and the existing cell labels.

143. The computer-implemented method of claim 142, further comprising generating a segmentation mask comprising the final cell labels.

144. A system comprising one or more computer processors and computer memory coupled thereto, the computer memory comprising a machine executable code that, upon execution by the one or more computer processors, implements the method of any one of the preceding claims.

145. A computer-implemented system comprising a computing device comprising at least one processor and instructions executable by the at least one processor to provide an application for improved cell segmentation in a biological sample, comprising: a) a software module configured to apply an analysis machine-learning algorithm for clustering, the analysis machine-learning algorithm configured to perform operations comprising: (i) obtaining a plurality of scanned images, wherein each scanned image of the plurality of scanned images comprises one or more cells, wherein the one or more cells are labeled with a marker; (ii) extracting, by a computer processor, pixel-level expression data of the marker of each scanned image of the plurality of scanned images, (iii) determining, by the computer processor, a degree of correlation between the pixel level expression data and the spatial distribution of the marker of each scanned image of the plurality of scanned images, and (iv) identifying, by the computer processor and based on the degree of correlation determined in (a)(iii), a plurality of clusters of the marker of each scanned image of the plurality of scanned images, wherein each cluster of the plurality of clusters comprises an identifier for a cell type, or a subcellular component, or a tissue characteristic; (b) a software module configured to normalizing the expression of each marker of each identifier of (a)(iv);WSGR Docket No.68840-711.601 (c) generating an aggregate intensity of expression of the markers for each cluster of the plurality of clusters identified in (a)(iv); (d) a software module configured to iteratively apply a segmentation machine-learning algorithm to each cluster of the plurality of clusters identified in (a)(iv), wherein the machine- learning model is configured to perform operations comprising: (i) assessing spatial flows and cell probability of the cluster, based on the aggregate intensity of expression of (c), and (ii) repeating the assessing for each cluster of the plurality of clusters; (e) a software module configured to combine data comprising the assessed spatial flows and cell probability of each cluster of the plurality of clusters; and (f) a software module configured to perform operations comprising generating a cell segmentation mask based at least in part on the data of (e).

146. The computer-implemented system of claim 145, wherein the analysis machine-learning algorithm comprises an unsupervised machine-learning algorithm.

147. The computer-implemented system of claim 145 or 146, wherein the analysis machine- learning algorithm comprises an unsupervised clustering algorithm.

148. The computer-implemented system of any one of claims 145-147, wherein the analysis machine-learning algorithm comprises one or more of a Louvian clustering algorithm, a Leiden clustering algorithm, or a K-means clustering algorithm, or any combination thereof.

149. The computer-implemented system of claim 148, wherein the segmentation machine- learning model comprises one or more types of segmentation machine-learning models.

150. The computer-implemented system of claim 149, wherein the operations further comprise applying a predetermined type of segmentation machine-learning model to each cluster of a subset of the plurality of clusters, wherein the subset of the plurality of clusters comprise the same identifier.

151. The computer-implemented method of claim 150, wherein the operations further comprise independently associating each identifier with a type of segmentation machine- learning model.

152. The computer-implemented system of claim 151, wherein the software module configured to apply the segmentation machine-learning model in (c) is further configured toWSGR Docket No.68840-711.601 perform operations comprising determining a cell boundary of each of the one or more cells of the biological sample.

153. The computer-implemented system of claim 152, wherein the software module configured to apply the segmentation machine-learning model in (c) is further configured to perform operations comprising determining a cell boundary based at least in part on the combined data comprising the assessed spatial flows and the cell probability of each cluster of the plurality of clusters.

154. The computer-implemented system of claim 145 or 153, wherein the software module configured to apply the segmentation machine-learning model in (c) is further configured to perform operations comprising generating the cell segmentation mask based at least in part on the cell boundary of the one or more cells of the biological sample.

155. The computer-implemented system of any one of claims 145-154, wherein the software module configured to apply the segmentation machine-learning model in (c) is further configured to perform operations comprising generating one or more cell labels based at least in part on the data of (d).

156. The computer-implemented system of claim 155, wherein the software module configured to apply the segmentation machine-learning model in (c) is further configured to perform operations comprising generating the cell segmentation mask based at least in part on the generated one or more cell labels.

157. The computer-implemented system of any one of claims 145-156, wherein the software module configured to apply the segmentation machine-learning model in (c) is further configured to perform operations comprising removing clusters with small area or low confidence in segmentation prediction.

158. The computer-implemented system of any one of claims 145-157, wherein the software module configured to apply the segmentation machine-learning model in (c) is further configured to perform operations comprising merging two different clusters based on the area of intersection region.

159. The computer-implemented system of any one of claim 145-158, wherein the software module configured to apply the segmentation machine-learning model in (c) is further configured to perform operations comprising splitting a cluster based on the area of intersection region.WSGR Docket No.68840-711.601 160. The computer-implemented system of any one of claims 145-159, wherein the intensity of expression data comprises proteomics data or transcriptomics data, or both.

161. The computer-implemented system of claim 160, wherein the proteomics data or transcriptomics data, or both comprise protein expression data, RNA expression data, or both.

162. The computer-implemented system of any one of claims 145-161, wherein the software module configured to apply the segmentation machine-learning model in (c) is further configured to perform operations comprising determining a degree of correlation between the intensity of expression and the spatial distribution of the plurality of clusters.

163. The computer-implemented system of claim 162, wherein the software module configured to apply the segmentation machine-learning model in (c) is further configured to perform operations comprising determining the aggregate intensity of expression and the spatial distribution of the plurality of clusters by analyzing transcriptomic data, proteomic data, or both.

164. The computer-implemented system of claim 163, wherein the transcriptomic data, proteomic data, or both comprise a human protein atlas database.

165. The computer-implemented system of any one of claims 145-164, wherein the segmentation machine-learning model comprises a cellular segmentation model.

166. The computer-implemented system of claim 165, wherein the cellular segmentation model comprises a Cellpose-style model, a Mesmer-style model, a StarDist model, or a deep watershed-based model.

167. The computer-implemented system of any one of claims 145-166, wherein the computer processor is further configured to apply the analysis machine learning algorithm and the segmentation machine learning model to each marker of a plurality of markers.

168. The computer-implemented system of claim 167, wherein the plurality of markers comprises at least about 5, at least about 10, at least about 20, at least about 30, at least about 50, at least about 100, at least about 200 or at least about 300 different markers.

169. The computer-implemented system of any one of claims 145-168, wherein the marker comprises a subcellular component.

170. The computer-implemented system of claim 169, wherein the subcellular component comprises a nucleus, a cell membrane, a cytoplasm, a soma, a protrusion, a mitochondria, a Golgi apparatus, a ribosome, an endoplasmic reticulum, a lysosome, a nucleolus, a nuclear speckles, a nucleoplasm, a nuclear periphery, a Cajal bodies, or a promyelocytic leukemia nuclear (PML) body.WSGR Docket No.68840-711.601 171. The computer-implemented system of any one of claims 145-170, wherein the biological sample or a derivative thereof is derived from brain, prostate, lung, colon, breast, ovary, uterus, esophagus, stomach, liver, kidney, skin, heart, muscle, or bone marrow.

172. The computer-implemented system of any one of claims 145-170, wherein the plurality of cells comprise one or more cell types selected from a neuron, an astrocyte, a microglial, an immune cell, an endothelial cell, an epithelial cell, a hepatocyte, a keratinocyte, a myoepithelial cell, or a muscle cell.

173. The computer-implemented system of any one of claims 145-172, the application further comprising generating, by the computer processor, one or more raw cell probability images.

174. The computer implemented system of any one of claims 145-173, the application further comprising assigning a value of zero to each existing cell label of the plurality of markers.

175. The computer implemented system of any one of claims 145-174, the application further comprising identifying one or more combined cell clusters.

176. The computer implemented system of claim 175, wherein the operations further comprise applying a threshold value for assessing spatial flows and cell probability of the one or more combined cell clusters.

177. The computer-implemented system of claim 176, wherein the threshold value is (1x10-1, max), (1x10-2, max), (1x10-3, max), (1x10-4, max), (1x10-5, max), (1x10-6, max), (1x10-7, max), (1x10-8, max), (1x10-9, max), (1x10-10, max), (1x10-11, max), (1x10-12, max), (1x10-13, max), (1x10-14, max), (1x10-15, max), or less than (1x10-15, max).

178. The computer-implemented system of claim 177, wherein the threshold value is (1x10-5, max).

179. The computer-implemented system of claim 178, the application further comprising applying a size-range filter using the threshold value.

180. The computer-implemented system of any one of claims 175-179, the application further comprising generating one or more new cell labels for the one or more combined cell clusters.

181. The computer-implemented of claim 180, wherein the one or more new cell labels comprise a marker.

182. The computer-implemented system any one of claims 175-181, wherein identifying one or more combined cell clusters comprises setting a threshold value of (0, max) or (-2, max) for assessing spatial flows and cellular probability of the one or more combined clusters.WSGR Docket No.68840-711.601 183. The computer-implemented system of any one of claims 175-182, the application further comprising applying a minimum size filter to each combined cluster of the one or more combined clusters.

184. The computer-implemented system of any one of claims 175-183, wherein the application further comprises generating the one or more new cell labels further comprises reassigning one or more of the existing cell labels.

185. The computer-implemented system of claim 184, wherein reassigning the one or more labels comprises determining the intersection of area (IoA) value of each combined cluster of the one or more combined clusters.

186. The computer-implemented system of claim 185, wherein the IoA comprises the touch IoA of each combined cluster of the one or more combined clusters.

187. The computer-implemented system of claim 184 or 185, wherein the application further comprises removing one or more of the labels from each combined cluster of the one or more combined clusters having an IoA threshold value of less than 0.

6.

188. The computer-implemented system of any one of claims 175-187, wherein the application further comprises selecting one or more of the existing labels to reassign by applying a nearest neighbor split algorithm.

189. The computer-implemented system of claim 188, wherein the nearest neighbor split algorithm is performed against the existing cell labels and the new cell labels.

190. The computer-implemented system of claims 188 or 189, wherein the nearest neighbor split algorithm is performed to generate one or more newly detected cell borders.

191. The computer-implemented system of claim 190, wherein the application further comprises generating the one or more new cell labels for one or more newly detected cells comprising the one or more newly detected cell borders.

192. The computer-implemented system of claim 191, wherein the application further comprises generating final cell labels by combining the new cell labels and the existing cell labels.

193. The computer-implemented system of claim 192, wherein the application further comprises generating a segmentation mask comprising the final cell labels.

Images