WO2015181371A1

WO2015181371A1 - An image processing method and system for analyzing a multi-channel image obtained from a biological tissue sample being stained by multiple stains

Info

Publication number: WO2015181371A1
Application number: PCT/EP2015/062015
Authority: WO
Inventors: Michael Barnes; Ting Chen; Alisa TUBBS; Carlo BIFULCO
Original assignee: Ventana Medical Systems, Inc.; F. Hoffmann-La Roche Ag; Providence Health & Services - Oregon
Priority date: 2014-05-30
Filing date: 2015-05-29
Publication date: 2015-12-03
Also published as: CA2944831A1; AU2015265811B2; US10275880B2; CA2944831C; US20170154420A1; JP2017529513A; JP6763781B2; EP3149700B1; AU2015265811A1; EP3149700A1

Abstract

Systems and methods for automatic FOV selection in immunoscore computation that involve reading images for individual markers from an unmixed multiplex slide or single stain slides, and computing the tissue region mask from the individual marker image. The heat map of each marker is determined by applying the low pass filter on the individual marker image channel and selecting the top K highest intensity regions from the heat map as the candidate FOVs for each marker. The candidate FOVs from the individual marker images are merged together in the same coordinate system by either adding all of the FOVs together or by only adding the FOVs from the selected marker images depending on the user's choice, and registering all the individual marker images to a common coordinate system and transferring the FOVs back to the original images.

Description

AN IMAGE PROCESSING METHOD AND SYSTEM FOR ANALYZING A MULTI-CHANNEL IMAGE OBTAINED FROM A BIOLOGICAL TISSUE

SAMPLE BEING STAINED BY MULTIPLE STAINS

BACKGROUND OF THE SUBJECT DISCLOSURE

Field of the Subject Disclosure

[0001] The present subject disclosure relates to imaging for medical diagnosis. More particularly, the present subject disclosure relates to automatic field of view (FOV) selection on a whole slide image.

Background of the Subject Disclosure

[0002] In the analysis of biological specimens such as tissue sections, blood, cell cultures and the like, biological specimens are stained with one or more combinations of stains to identify, for example, biomarkers, cells or cellular structures, and the resulting assay is viewed or imaged for further analysis. Observing the assay enables a variety of processes, including diagnosis of disease, assessment of response to treatment, and development of new drugs to fight disease. An assay includes one or more stains conjugated to an antibody that binds to protein, protein fragments, or other objects of interest in the specimen, hereinafter referred to as targets or target objects. Some biomarkers, for example, have a fixed relationship to a stain (e.g., the often used counterstain hematoxylin), whereas for other biomarkers, a stain may developed or anew assay may be created. Subsequent to staining, the assay may be imaged for further analysis of the contents of the tissue specimen. An image of an entire slide is typically referred to as a whole-slide image, or simply whole-slide.

[0003] Typically, in immunoscore computations, a scientist uses a multiplex assay that involves staining one piece of tissue or a simplex assay that involves staining adjacent serial tissue sections to detect or quantify, for example, multiple proteins or nucleic acids etc. in the same tissue block. With the stained slides available, the immunological data, for instance, the type, density and location of the immune cells, can be estimated from the tumor tissue samples. It has been reported that this data can be used to predict the patient survival of colorectal cancer and demonstrates important prognostic role.

[0004] In the traditional workflow for immunoscore computation, the expert reader such as a pathologist or biologist selects the representative fields of view (FOVs) or regions of interest (ROIs) manually, as the initial step, by reviewing the slide under a microscope or reading an image of a slide, which has been scanned / digitized, on a display. When the tissue slide is scanned, the scanned image is viewed by independent readers and the FOVs are manually marked based on the readers' personal preferences. After selecting the FOVs, a pathologist/reader manually counts the immune cells within the selected FOVs. Manual selection of the FOVs and counting is highly subjective and biased to the readers, as different readers may select different FOVs to count. Hence, an immunoscore study is no longer reproducible.

[0005] EP2546802 features an artificial hyper-spectral image generated from co- registered tissue slides that enables the sophisticated co-analysis of image stacks. Co-registration is performed on tiles of high-resolution images of tissue slices, and image-object statistics are used to generate pixels of a down-scaled hyper-spectral image. The method of analyzing digital images to generate hyperspectral images combines two hyperspectral images to generate a third hyperspectral image.

[0006] As another example, U.S. Publication No. 2014/0180977 features classifying histological tissues or specimens with two phases. In a first phase, the method includes providing off-line training using a processor during which one or more classifiers are trained based on examples. In a second phase, the method includes applying the classifiers to an unknown tissue sample with extracting the first set of features for all tissue units; deciding for which tissue unit to extract the next set of features by finding the tissue unit for which a score is maximized; iterating until a stopping criterion is met or no more feature can be computed; and issuing a tissue-level decision based on a current state. [0007] In Schonmeyer, an automated workflow where quantitative image analysis results of consecutive differently stained tissue sections are locally fused by co- registration. The results are spatially resolved feature vectors containing features which are hyperspectral. Heat maps with many layers (hyperspectral) are generated from this data, revealing relationships between different stains that would not be evident from single stains alone.

[0008] In Misemer, digital image analysis was performed to evaluate inter- and intratumor heterogeneity, and correlate protein expression with histologic features, including a histopathologic assessment of tumor activity, defined by nuclear chromatin density ratio (CDR). Pathologic assessment of tumor activity and digital assessment of average nuclear size and CDR were all significantly correlated.

SUMMARY OF THE SUBJECT DISCLOSURE

The invention relates to an image processing method for analyzing a multichannel image obtained from a biological tissue sample being stained by multiple stains and a respective image processing system as claimed in independent claims 1 and 9. Further embodiments are given in the dependent claims and further aspects of the invention are given in the further independent claims.

[0009] A 'biological tissue sample' as understood herein is any biological sample, such as a surgical specimen that is obtained from a human or animal body for anatomic pathology. The biological sample may be a prostrate tissue sample, a breast tissue sample, a colon tissue sample or a tissue sample obtained from another organ or body region.

[0010] A 'multi-channel image' as understood herein encompasses a digital image obtained from a biological tissue sample in which different biological structures, such as nuclei and tissue structures, are simultaneously stained with specific fluorescent dyes, each of which fluoresces in a different spectral band thus constituting one of the channels of the multi-channel image. The biological tissue sample may be stained by a plurality of stains and/or by a stain and a counterstain, the later being also referred to as a "single marker image".

[0011] An 'unmixed image' as understood herein encompasses a grey-value or scalar image obtained for one channel of a multi-channel image. By unmixing a multichannel image one unmixed image per channel is obtained.

[0012] A 'color channel' as understood herein is a channel of an image sensor. For example, the image sensor may have three color channels, such as red (R), green (G) and blue (B).

[0013] A 'heat map' as understood herein is a graphical representation of data where the individual values contained in a matrix are represented as colors.

[0014] 'Thresholding' as understood herein encompasses the application of a

predefined threshold or sorting of local maxima to provide a sorted list and selecting of a predetermined number of the local maxima from the top of the sorted list.

[0015] 'Spatial low pass filtering' as understood herein encompasses a spatial filtering using a spatial filter that performs a low pass filtering operation on a neighborhood of image pixels, in particular a linear or non-linear operation. In particular, spatial low pass filtering may be performed by applying a convolutional filter. Spatial filtering is as such known from the prior art, (cf. Digital Image Processing, Third Edition, Rafael C. Gonzalez, Richard E. Woods, page 145, chapter 3.4.1 ).

[0016] 'Local maximum filtering' as understood herein encompasses a filtering operation where a pixel is considered a local maximum if it is equal to the maximum value in a subimage area. Local maximum filtering can be implemented by applying a so called max filter, (cf. Digital Image Processing, Third Edition, Rafael C. Gonzalez, Richard E. Woods, page 326, chapter 5).

[0017] A 'field of view (FOV)' as understood herein encompasses an image portion that has a predetermined size and shape, such as a rectangular or circular shape.

[0018] Embodiments of the present invention are particularly advantageous as an

automatic and reliable technique is provided to identify fields of view in a multichannel image while avoiding the tedious effort of manually marking fields of view in a multi-channel image by a pathologist or biologist and thereby also eliminating subjective judgment and human error. As the spatial low pass filtering, the local maximum filtering and thresholding operations can be executed at high processing speeds, the computational expense and the latency time experienced by the user can be minimized. This is due to the fact that the definition of the fields of view is not performed directly on the multi-channel image but on the basis of the filtered and thresholded image which enables the high processing speed.

[0019] It is to be noted that the analysis in step f is executed on the full resolution multichannel image and not on the spatial low pass filtered unmixed image. This assures that the full amount of the available pictorial information can be used for performing the analysis while the filtering operation, namely steps b, c and d, merely serve for identification of the relevant fields of view where a full analysis is to be performed.

[0020] In accordance with a further embodiment of the invention one of the unmixed images is processed for defining the field of view as described above while another one of the unmixed images is segmented for identification of tissue regions. Suitable segmentation techniques are as such known from the prior art, (cf. Digital Image Processing, Third Edition, Rafael C. Gonzalez, Richard E. Woods, chapter 10, page 689 and Handbook of Medical Imaging, Processing and Analysis, Isaac N. Bankman, Academic Press, 2000, chapter 2). By means of the segmentation non-tissue regions are removed as the non-tissue regions are not of interest for the analysis.

[0021] The segmentation provides a mask by which those non-tissue regions are

removed. The resultant tissue mask can be applied onto the unmixed image prior or after the spatial low pass or local maximum filtering or thresholding operations and before or after the fields of view are defined. It may be advantageous to apply the tissue mask at an early stage in order to further reduce the processing load, such as before the execution of the spatial low pass filtering.

[0022] In accordance with an embodiment of the invention the other one of the unmixed images that is segmented for providing the tissue mask is obtained from the channel that is representative of one stain that is a counter-stain to the stain represented by the unmixed image that is processed in accordance with steps b-e of claim 1.

[0023] In accordance with an embodiment of the invention fields of view are defined for at least two of the unmixed images. Fields of view that are defined in two different unmixed images can be merged if they are located at the same or almost identical image location. This is particularly advantageous for stains that can be co-located such that a single field of view results for the co-located stains that identify a common biological structure. By merging such fields of view the processing load is further reduced and the analysis in step f needs only to be performed once for the merged field of view. Moreover, the cognitive burden for the pathologist or biologist is also reduced as only one analysis result is presented rather than two related results. Depending on the implementation, the two fields of view may be merged if a degree of spatial overlap of the fields of view is above an overlap threshold.

[0024] In accordance with embodiments of the invention the analysis of the field of view is performed by cell counting of the biological cells shown in the multi-channel image within the considered field of view. The cell counting can be performed by using a suitable image analysis technique which is applied on the field of view. In particular, the cell counting can be executed by means of an image classification technique.

[0025] In accordance with further embodiments of the invention the analysis of the field of view is performed by means of a trained convolutional neural network such as by entering the field of view or an image patch taken from the field of view into the convolutional neural network for determining a probability for the presence of a biological feature within the field of view or the image patch, respectively. An image patch may be extracted from the field of view for entry into the convolutional neural network by first identifying a location of interest within the field of view and then extracting the image patch that contains this location of interest.

[0026] In accordance with a further embodiment of the invention the analysis is

performed on the field of view in step f as a data analysis, such as a cluster analysis or statistical analysis..

[0027] In accordance with another aspect of the invention an image processing system for analyzing a multi-channel image obtained from a biological tissue sample being stained by multiple stains is provided that is configured to execute a method of the invention. [0028] The subject disclosure provides systems and methods for automatic field of view (FOV) selection based on a density of each cell marker in a whole slide image. Operations described herein include reading images for individual markers from an unmixed multiplex slide or from singularly stained slides, and computing the tissue region mask from the individual marker image. A heat map of each marker may be determined by applying a low pass filter on an individual marker image channel, and selecting the top K highest intensity regions from the heat map as the candidate FOVs for each marker. The candidate FOVs from the individual marker images are merged together. The merging may comprise one or both of adding all of the FOVs together in the same coordinate system, or only adding the FOVs from the selected marker images, based on an input preference or choice, by first registering all the individual marker images to a common coordinate system and merging through morphologic operations. After that, all of the identified FOVs are transferred back to the original images using inverse registration to obtain the corresponding FOV image at high resolution. Without wishing to limit the present invention to any theory or mechanism, the systems and methods of the present invention may offer advantages such as being reproducible, unbiased to human readers, and more efficient.

[0029] The present methodology of automated selection of the FOVs using a plurality of low-resolution single image marker slides greatly improves the reliability and efficiency of the FOV selection process. By automating the selection of the fields of view, a uniform method is applied reducing the subjectivity of independent readers. Use of low-resolution images to perform the FOV selection furthermore improves computational efficiency, allowing the analyst to rapidly proceed to analysis of the tissue regions. In some embodiments, the subject disclosure uses lower-resolution images to speed computation of the FOVs. Because the images are lower resolution, it is computationally much faster to compute the heat map and tissue region mask. This allows the selection of the FOVs to be made automatic and rapid, which allows for faster analysis of the tissue sample. [0030] In some embodiments, the subject disclosure provides a system for quality control of automated whole-slide analysis, including a processor, and a memory coupled to the processor, the memory to store computer-readable instructions that, when executed by the processor, cause the processor to perform operations comprising merging a portion of a plurality of candidate fields of view (FOVs) of an image of an IHC slide, and depicting the merged portion of the plurality of candidate fields of view on the image.

[0031] In another embodiment, the present invention features a system for quality control of automated detection of structures in an image of a biological tissue sample. An image acquisition system, a processor, and a memory coupled to the processor may be included in the system. The memory can store computer- readable instructions that, when executed by the processor, cause the processor to perform operations such as receiving a color image from the image acquisition system, applying a color unmixing operation to the image to produce a single- color image in a blended color channel of interest, applying a spatial frequency filter to the single-color image, applying a local max filter to the filtered color channel image, thresholding the local max filter image to produce a binary image that contains a region of interest, and identifying an isolated binary region as a candidate field of view (FOV).

[0032] In some embodiments, the subject disclosure provides a tangible non- transitory computer-readable medium to store computer-readable code that is executed by a processor to perform operations including computing a tissue region mask from each of a plurality of images associated with a patient, generating a list comprising a plurality of candidate fields of view for each of the plurality of images based on the tissue region mask, merging a portion of the plurality of candidate fields of view (FOVs), and depicting the merged portion of the plurality of candidate fields of view on one or more of the plurality of images.

[0033] In some embodiments, the subject disclosure provides a tangible non- transitory computer-readable medium to store computer-readable code that is executed by a processor to perform operations including loading a list of image folders, wherein each folder contains a plurality of images corresponding to a plurality of biomarkers associated with a patient, displaying heat maps for each of the biomarkers identified in the images, and allowing a user to select a number of FOVs from at least one of the images or heat maps.

[0034] In another embodiment, the present invention features a tangible, non- transitory computer-readable medium having computer-executable instructions for execution by a processing system, the computer-executable instructions for automatic detection of structures in an image of a biological tissue sample. The computer-executable instructions, when executed, can cause the processing system to perform operations such as receiving a color image from the image acquisition system, applying a color unmixing operation to the image to produce a single-color image in a blended color channel of interest, applying a spatial frequency filter to the single-color image, applying a local max filter to the filtered color channel image, thresholding the local max filter image to produce a binary image that contains a region of interest, and identifying an isolated binary region as a candidate field of view (FOV).

[0035] Another embodiment of the subject disclosure features a method for automatic detection of structures in an image of a biological tissue sample. The method may be implemented by an imaging system, and may be stored on a medium readable by a computing device. These methods may comprise logical instructions that are executed by a processor to perform operations such as receiving a color image, applying a color unmixing operation to the image to produce a single-color image (which is also referred to as scalar value or grey- scale image) , applying a spatial frequency filter to the single-color image such that the spatial frequencies corresponding to structural features of interest are enhanced and all other spatial frequencies are diminished, applying a local max filter to the filtered color channel image such that the local max filter kernel has spatial extent on the order of the spatial extent of features enhanced by the spatial filtering applied to the single-color image, thresholding the local max filter image to produce a binary image that contains a region of interest, and identifying an isolated binary region as a candidate field of view (FOV). In some embodiments, the binary image may identify a plurality of regional maxima in the image, and then further identify a plurality of candidate FOVs based on the identified regional maxima. In other embodiments, the plurality of candidate FOVs are associated with an intensity value relative to the other intensity values in the image.

[0036] In still other embodiments, the operations may further comprise ranking the plurality of candidate FOVs in ascending or descending order, receiving and outputting a predetermined number of the plurality of candidate FOVs, and selecting the predetermined number of FOVs from the highest intensity regions.

[0037] In addition, the present invention is surprisingly effective to rapidly collect significant diagnostic indicators, and allows diagnostic workers to rapidly scan a pre-selected collection of diagnostically important regions. Without limiting the invention to any particular operation or mechanism, it is believed that the step of applying a spatial frequency filter, such as a Gaussian filter or a local averaging filter, to the single-color image (wherein the spatial frequencies corresponding to structural features of interest are enhanced, and all other spatial frequencies are diminished) prior to the step of applying a local max filter to the filtered color channel image, enables this rapid collection of significant diagnostic indicators. Prior art relating to computer implemented collection of significant diagnostic indicators on histology slides do not have these steps.

[0038] Tissue slide images contain many features, only some of which are of interest for any particular study. Those interesting regions may have a specific color brought about by selective stain uptake. They may also have broad spatial extent. Importantly, the uninteresting regions may have some specific spatial frequencies that enable their removal from an image by way of spatial frequency filtering. Such filters include, but are not limited to, low pass, high pass, and band pass, filters. More carefully tuned spatial frequency filters may be those known as matched filters. Non-limiting examples of spatial frequency filters include, but are not limited to, low pass filters, high-pass filters, band-pass filters, multiple- passband filters, and matched filters. Such filters may be statically defined, or adaptively generated.

[0039] In the process of locating regions of interest, it is therefore helpful to first select the proper color by an unmixing process, which can be viewed as a linear operator applied to the primary color channels, R, G, and B, of the image. Spatial frequency filtering is also applied to give preference to features of interest in the image. These operations may be applied in either order since they are both linear operators.

[0040] In parallel with this region selection, there may be a broader segmentation mask formed by using entirely differently tuned spatial frequency filters, to select, for example, only the gross region of the slide image where tissue resides, and rejecting empty regions. Therefore, multiple different spatial frequency filters may be applied to the same tissue slide image.

[0041] Once filtered, a region of interest may be located by applying a local max filter, a kind of morphological nonlinear filter, which produces an image by making each pixel of the result hold the value of the maximum pixel value from the source image that lies beneath the kernel of the max filter. The kernel is a geometric mask of arbitrary shape and size, but would be constructed for this purpose to have dimensions on the order of the interesting features. The output image from a local max filter will tend to have islands shaped like the kernel and with constant values equal to the maximum pixel value in that region.

[0042] In some embodiments, with the present construction of a local max filter image, a threshold may be applied to convert the filter image to a binary mask, by assigning binary mask values of 1 to corresponding filter image pixels above the threshold, and values of 0 to corresponding filter image pixels below the threshold. The result will be blobs of 1 's that can be labeled as regions, and with measureable spatial extents. Together, these region labels, locations, and spatial extents provide a record of regions of interest (ROIs), or fields of view (FOVs).

BRIEF DESCRIPTION OF THE DRAWINGS

[0043] FIGS. 1A-1 B respectively depict a system and a workflow for automatic FOV selection, according to an exemplary embodiment of the present subject disclosure.

[0044] FIG. 2 depicts a heat map computation, according to an exemplary embodiment of the present subject disclosure.

[0045] FIG. 3 depicts a tissue mask computation, according to an exemplary embodiment of the subject disclosure.

[0046] FIG. 4 depicts candidate FOVs, according to an exemplary embodiment of the subject disclosure.

[0047] FIGS. 5A-5B depict merging of FOVs from all markers and from selected markers, respectively, according to an exemplary embodiment of the subject disclosure.

[0048] FIGS. 6A-6B depict integrating FOVs, according to exemplary embodiments of the subject disclosure.

[0049] FIG. 7 depicts a user interface for image analysis using an all marker view, according to an exemplary embodiment of the subject disclosure.

[0050] FIG. 8 depicts a user interface for image analysis using an individual marker view, according to an exemplary embodiment of the subject disclosure.

[0051] FIG. 9 depicts a digital pathology workflow for immunoscore computation, according to an exemplary embodiment of the subject disclosure.

[0052] FIG. 10 depicts a process flow chart for an exemplary embodiment of the present invention of the present invention.

[0053] FIGs. 1 1a and 1 1 b depicts a process flow chart for an exemplary embodiment of the present invention starting with single-stain marker images.

[0054] FIG. 12 depicts a process flow chart for an exemplary embodiment of the present invention starting with a multiplex slide.

[0055] FIG. 13 depicts a process flow chart for an exemplary embodiment of the present invention starting with a single stain image.

DETAILED DESCRIPTION OF THE SUBJECT DISCLOSURE

[0056] The present invention features systems and methods for automatic field of view (FOV) selection based on a density of each cell marker in a whole slide image. Operations described herein include but are not limited to reading images for individual markers from an unmixed multiplex slide or from singularly stained slides, and computing the tissue region mask from the individual marker image. A heat map of each marker may be determined by applying a low pass filter on an individual marker image channel, and selecting the top K highest intensity regions from the heat map as the candidate FOVs for each marker. The candidate FOVs from the individual marker images may then be merged together. The merging may comprise one or both of adding all of the FOVs together in the same coordinate system, or only adding the FOVs from the selected marker images, based on an input preference or choice, by first registering all the individual marker images to a common coordinate system and merging through morphologic operations. Subsequently, all of the identified FOVs are transferred back to the original images using inverse registration to obtain the corresponding FOV image at high resolution. Without wishing to limit the present invention to any theory or mechanism, the systems and methods of the present invention may offer advantages such as being reproducible, unbiased to human readers, and more efficient.

[0057] In some embodiments, the system for quality control of automated whole- slide analysis comprises an image acquisition system (102), a processor (105); and a memory coupled to the processor (1 10). The memory is configured to store computer-readable instructions that, when executed by the processor, cause the processor to perform operations one or more of the following operations (but not limited to the following operations) comprising: reading an image, for example, a high resolution input image (231 ) from the image acquisition system (102), computing or receiving a low resolution version of the high resolution input image, reading a plurality of low resolution image marker images from the image acquisition system (102), wherein each image marker image is of a single color channel (232) of the low resolution input image, computing a tissue region mask (233) corresponding to the low resolution input image, computing a low pass filtered image (234) of each image marker image (1 14), generating a masked filtered for each image marker image (1 13), where the masked filtered image is the tissue region mask multiplied by the low pass filtered image, identifying a plurality of candidate fields of view (FOVs) within each masked filtered image (1 16), merging a subset of a plurality of candidate FOVs for each image marker image (1 17), into a plurality of merged FOVs, and depicting the merged portion of the plurality of candidate fields of view on the input image.

[0058] In some embodiments, a heat map may be computed for the masked filtered image. In some embodiments, the heat map comprises applying colors to the masked filtered image, wherein low intensity regions are assigned to blue colors and higher intensity regions are assigned to yellow orange and red colors. Any other appropriate colors or combinations of colors may be used to assign low and high intensity regions.

[0059] In some embodiments, the generation of the tissue region mask comprises one or more of the following operations (but not limited to the following operations): computing the luminance (337) of the low resolution input image(336), producing a luminance image (338), applying a standard deviation filter to the luminance image (339), producing a filtered luminance image (340), and applying a threshold to filtered luminance image (341 ), such that pixels with a luminance above a given threshold are set to one, and pixels below the threshold are set to zero, producing the tissue region mask (342). [0060] In some embodiments, the tissue region mask is computed directly from the high resolution input image. In this case, the tissue region mask may be converted to a lower resolution image before application to the filtered image market images.

[0061] In some embodiments, the image marker images are obtained by unmixing (1 1 1 ) a multiplex slide, where the unmixing module uses a reference color matrix (1 12) to determine what colors correspond to the individual color channels. In other embodiments, the image marker images are obtained from single stain slides.

[0062] In some embodiments, the image registration process comprises selecting one image marker image to serve as a reference image, and computing a transformation of each image marker to the coordinate frame of the reference image. The methods for computing a transformation of each image to a reference image are well known to those skilled in the art. In other embodiments, if the images are obtained by unmixing a multiplex reference slide, no registration is needed since all the unmixed images are already in the same coordinate system.

[0063] The subject disclosure provides systems and methods for automatic field of view (FOV) selection. In some embodiments, the FOV selection is based on a density of each cell marker in a whole slide image. Operations described herein include reading images for individual markers from an unmixed multiplex slide or from singularly stained slides, and computing the tissue region mask from the individual marker image. A masked filtered image of each marker may be determined by applying a low pass filter on an individual marker image channel, and applying the tissue region mask. The top K highest intensity regions from the masked filtered image are selected as the candidate FOVs for each marker. The candidate FOVs from the individual marker images are merged together. The merging may comprise one or both of adding all of the FOVs together in the same coordinate system, or only adding the FOVs from the selected marker images, based on an input preference or choice, by first registering all the individual marker images to a common coordinate system and merging through morphologic operations. After that, all of the identified FOVs are transferred back to the original images using inverse registration to obtain the corresponding FOV image at high resolution. Without wishing to limit the present invention to any theory or mechanism, the systems and methods of the present invention may offer advantages such as being reproducible, unbiased to human readers, and more efficient. As a result, a digital pathology workflow for automatic FOV selection, in accordance with the subject disclosure, includes a computer-based FOV selection algorithm that automatically provides the candidate FOVs that may be further analyzed by a pathologist or other evaluator.

[0064] The operations described herein have been described, for exemplary purposes, in connection with the identification of immune cells, and for use in immunoscore computations. However, the systems and methods may be applicable to any type of image of a cell or biological specimen, and are applicable to determinations of type, density and location for any type of cell or group of cells. As used herein, the terms "biological specimen" and "biological tissue sample" may be used interchangeably. Moreover, besides cancerous tissue and immune markers, the subject disclosure is applicable to any biological specimen or tumor of any disease or non-disease state, and images of biological specimens that have been subjected to any type of staining, such as images of biological specimens that have been stained with fluorescent and non-fluorescent stains. Also, one of ordinary skill in the art would recognize that the order of the steps may vary from what is described herein.

[0065] FIGS. 1A-1 B respectively depict a system 100 and a workflow for automatic FOV selection, according to an exemplary embodiment of the present subject disclosure. Referring to FIG. 1A, a system 100 comprises a memory 1 10, which stores a plurality of processing modules or logical instructions that are executed by processor 105 coupled to computer 101. An input from image acquisition system 102 may trigger the execution of one or more of the plurality of processing modules. Image acquisition system 102 may comprise an image sensor that has color channels such as RGB. Besides processor 105 and memory 1 10, computer 101 also includes user input and output devices such as a keyboard, mouse, stylus, and a display / touchscreen. As will be explained in the following discussion, processor 105 executes logical instructions stored on memory 1 10, including automatically identifying one or more FOVs in an image of a slide (containing a biological specimen, such as a tissue sample) that has been stained with one or more stains (for example, fluorophores, quantum dots, reagents, tyramides, DAPI, etc.).

[0066] Image acquisition system 102 may include a detector system, such as a CCD detection system having a CCD image sensor, or a scanner or camera such as a spectral camera, or a camera on a microscope or a whole-slide scanner having a microscope and/or imaging components (the image acquisition system is not limited to the aforementioned examples). For example, a scanner may scan the biological specimen (which may be placed on a substrate such as a slide), and the image may be saved in a memory of the system as a digitized image. Input information received from image acquisition system 102 may include information about a target tissue type or object, as well as an identification of a staining and/or imaging platform. For instance, the sample may have been stained by means of application of a staining assay containing one or more different biomarkers associated with chromogenic stains for brightfield imaging or fluorophores for fluorescence imaging. Staining assays can use chromogenic stains for brightfield imaging, organic fluorophores, quantum dots, or organic fluorophores together with quantum dots for fluorescence imaging, or any other combination of stains, biomarkers, and viewing or imaging devices. Moreover, a typical sample is processed in an automated staining/assay platform that applies a staining assay to the sample, resulting in a stained sample. Input information may further include which and how many specific antibody molecules bind to certain binding sites or targets on the tissue, such as a tumor marker or a biomarker of specific immune cells. The choice of biomarkers and/or targets may be input into the system, enabling a determination of an optimal combination of stains to be applied to the assay. Additional information input into system 100 may include any information related to the staining platform, including a concentration of chemicals used in staining, a reaction times for chemicals applied to the tissue in staining, and/or pre-analytic conditions of the tissue, such as a tissue age, a fixation method, a duration, how the sample was embedded, cut, etc. Image data and other input information may be transmitted directly or may be provided via a network, or via a user operating computer 101.

[0067] An unmixing module 1 1 1 may be executed to unmix the image, for instance if the image is a multiplex image. Unmixing module 1 1 1 unmixes the image into individual marker color channels. Unmixing module 1 1 1 may read from a reference color matrix database 1 12 to obtain the reference color matrix and use the reference color matrix to perform unmixing operations. If the image is of a single stain slide, the image can be directly used for FOV selection. In either case, a heat map computation module 1 13 may be executed to evaluate a heat map for each individual marker image, or single stain image. A heat map maps the density of various structures or biomarkers on the whole-slide image. To accomplish this, heat map computation module 1 13 may perform operations such as assigning colors to a low pass filtered image that is processed by low pass filter module 1 14. A tissue region mask may also be applied to the low pass filtered image. The heat map illustrates pixels according to the respective densities of the pixels, and thus, corresponds to the density of the cell distribution in each image. For example, the heat map will distinguish high-density pixels from low-density pixels by illustrating higher density pixels in a color that is warmer than a color used for lower density pixels, where a 'high-density pixel' refers to a pixel that has a high pixel value. Local max filter module 1 15 may be executed to apply a local max filter to the low pass filtered image to obtain the local maxima of the image. Subsequently, a top K FOV selection module 1 16 may be executed to select the top K regions with the highest densities from the local max filtered image. The top K regions are designated as the candidate FOVs for each image. For example, the cells may be clustered together in the high-density region while they are more scattered in the low-density region. The FOVs from each image are merged together by merge FOV module 1 17, which performs operations such as taking all the FOVs or the FOVs from selected markers only and merging them. A registration module 1 18 is invoked to transfer all the images to the same coordinate system, so that the coordinates of the FOVs can be directly added up in the same coordinate system.

[0068] As described above, the modules include logic that is executed by processor 105. "Logic", as used herein and throughout this disclosure, refers to any information having the form of instruction signals and/or data that may be applied to affect the operation of a processor. Software is one example of such logic. Examples of processors are computer processors (processing units), microprocessors, digital signal processors, controllers and microcontrollers, etc. Logic may be formed from signals stored on a computer-readable medium such as memory 1 10 that, in an exemplary embodiment, may be a random access memory (RAM), read-only memories (ROM), erasable / electrically erasable programmable read-only memories (EPROMS/EEPROMS), flash memories, etc. Logic may also comprise digital and/or analog hardware circuits, for example, hardware circuits comprising logical AND, OR, XOR, NAND, NOR, and other logical operations. Logic may be formed from combinations of software and hardware. On a network, logic may be programmed on a server, or a complex of servers. A particular logic unit is not limited to a single logical location on the network. Moreover, the modules need not be executed in any specific order. Each module may call another module when needed to be executed.

[0069] An exemplary workflow for FOV selection is depicted in FIG. 1 B. In FIG. 1 B, N represents the number of markers applied to the slides. For a multiplex slide 121 , color unmixing 122 is performed, for example according to the unmixing method disclosed in Patent Application 61/830,620, filed June 3, 2013, and WO 2014/195193 A1 entitled "Image Adaptive Physiologically Plausible Color Separation", the disclosure of which is hereby incorporated by reference in its entirety. The method disclosed in Patent Application 61/943,265, filed February 21 , 2014, and entitled. "Group Sparsity Model for Image Unmixing", and PCT/EP2014/078392 filed 18 December 2014 [0070] which are hereby incorporated by reference in their entirety, disclose exemplary embodiments utilized to obtain an image 123 for each marker. Otherwise, if the image is a single stain slide, scanned images 124 of single stain slides for each marker are utilized as an input to an automatic FOV selection system, such as the system depicted in FIG. 1A. For example, a heat map computation operation may be performed to compute the hotspot 125 from the image of each marker to generate the top candidate FOVs 126 for each marker. The candidate FOVs 126 may be integrated 127 to generate the final FOV list 128. Final FOV list 128 comprises a list of possible FOVs for selection by a pathologist to utilize for evaluating the biological specimen, for example, immune cells.

[0071] As used herein and throughout this disclosure, hotspots are regions containing a high density of marked (i.e., stained) cells, for example hotspots can be cells from different types of images and markers such as In Situ Hybridization (ISH) markers, immunohistochemistry (IHC) markers, fluorescent markers, quantum dots etc. The subject disclosure uses immune cells in an IHC image as an example to demonstrate this feature (as previously discussed, the present invention is not limited to immune cells in an IHC image). In light of the subject disclosure, various algorithms may be used by those having ordinary skill in the art to find hotspots and to use automatic hotspot selection as a module in immunoscore computation. Exemplary embodiments of the subject disclosure utilize the automatic FOV selection operations described herein to solve the problem of avoiding biased manually selected FOVs. To automatically identify FOVs that may be of interest to a pathologist or other evaluator, a heat map is computed for each marker or image representing a single marker, based on a low-resolution image (e.g. a 5x zoom image).

[0072] FIG. 2 depicts a heat map computation, according to an exemplary embodiment of the present subject disclosure. The operations described in FIG. 2 illustrate how a heat map computation is utilized to identify hotspots. For example, given a single-marker channel 232 of an input image 231 , a low-pass- filtered image 234 is used to generate heat map 235, which basically takes the low pass filtered image 234 as input and applies a color map on top of it for visualization purposes. For example, a red color may correspond to high intensity pixels in the low pass filtered image and a blue color may correspond to low intensity pixels. Other depictions of color and/or intensity may be evident to those having ordinary skill in the art in light of this disclosure. A tissue region mask 233 may be created by identifying the tissue regions and excluding the background regions. This identification may be enabled by image analysis operations such as edge detection, etc. Tissue region mask 233 is used to remove the non-tissue background noise in the image, for example the non- tissue regions.

[0073] In the embodiment considered with respect to Fig. 2 the input multichannel image 231 is stained by means of a stain and its respective counter- stain which provides two channels, namely the FP3 channel and the HTX channel. The multi-channel image 231 is unmixed which provides the unmixed images 232 and 238 of the FP3 and HTX channels, respectively.

[0074] The unmixed image 232 is then low pass filtered by means of a spatial low pass filter which provides the low pass filtered image 234. Next, the heat map 235 may be added to the low pass filtered image 234 for visualization purposes.

[0075] The low pass filtered image 234 with or without the added heat map 235 is then local maximum filtered which provides the local max filtered image 236. The local max filtered image 236 comprises a number of local maxima 239, in the example considered here five local maxima 239.1 -239.5 as depicted in FIG. 2. Next, a thresholding operation is performed on the local max filtered image 236 such as by applying a threshold onto the local max filtered image 236 such that only the local maxima 239.1 and 239.4 that surpass this threshold are not removed by the thresholding operation.

[0076] Alternatively the local maxima 239 are ranked in a sorted list and only a number of the K topmost local maxima are taken from the list, where K is 2 for explanatory purposes in the embodiment considered here, resulting in the local maxima 239.1 and 239.4. Each of the local maxima 239 consists of a set of neighboring pixels.

[0077] This thresholding operation provides the thresholded image 237.

[0078] Each of the local maxima 239.1 and 239.4 in the thresholded image 237 may define the location of a respective field of view 240.1 and 240.2, respectively. Depending on the implementation, these fields of view 240.1 and 240.2 may be candidate fields of view for testing whether these fields of view can be merged with other fields of view in subsequent processing operations as described below with respect to FIG. 6. The positions of the fields of view 240.1 and 240.2 are defined by means of the thresholded image 237 and its local maxima. However, the content of the fields of view is taken from the respective image area within the original multi-channel image 231 in order to take advantage of the full pictorial information content for performing an image analysis of the respective field of view.

[0079] FIG. 3 depicts a tissue mask computation, according to an exemplary embodiment of the subject disclosure, such as to compute tissue mask 233 by means of a segmentation technique. A linear combination 337 of the RGB channels 336 of the tissue RGB image is computed to create a grayscale luminance image 338. The combination weights for the R, G and B channels (e.g. 0.3, 0.6, 0.1 in 337) are subject to change based on different applications. A 3 pixel by 3 pixel standard deviation filter 339 is applied to the luminance image 338, resulting in a filtered luminance image 340. Here the filter size (e.g. 3 by 3, 5 by 5) is subject to change based on different applications. The tissue mask 342 is a binary image obtained from thresholding 341 the filtered luminance image 340. For example, tissue mask 342 may comprise regions with pixel intensity value larger than 1.5. The thresholding parameter MaxLum (e.g. 1.5, 2.0, 3.0) can vary based on different applications.

[0080] FIG. 4 depicts candidate FOVs, according to an exemplary embodiment of the subject disclosure. Candidate FOVs 443 are selected from the top K highest density regions (also called hot spots) of the heat map. For example, K can be chosen from 5, 10, 15, 20 etc.

[0081] A local maximum filter is applied to the low pass filtered image 234 with the added heat map 235 (cf. Fig. 2) in order to provide a local max filtered image 236 It is to be noted that the heat map 235 is not essential for the processing but serves for visualization purposes.A local maximum filter is a function to identify a constant value connected region of pixels with the external boundary pixels all having a lower value. It can use 4 or 8 connected neighborhoods for 2-D images. The implementation of this functionality is available at Matlab (http://www.mathworks.com/help/images/ref/imregionalmax.html). The local maximum is obtained as the average intensity with in the connected region. The local maximum values are sorted providing a sorted list to produce the rank of the hotspots and top K hotspots are reported thus thresholding the local max filtered image. Alternatively a predefined threshold is applied on the local maximum filtered image such that all hotspots above the threshold are reported. The regions returned by the local maximum filter computation module are the locations of the local maximums.

[0082] As described herein, different FOVs may be obtained for different marker images resulting from unmixing of a multiplex slide or from single stain slides. The FOVs are integrated to ensure that for each patient under diagnosis, the same set of FOVs is referenced across different markers. There are several possible options to integrate FOVs. FIGS. 5A-5B depict merging of FOVs from all markers and from selected markers, respectively, according to an exemplary embodiment of the subject disclosure. For example, all candidate FOVs from the different marker images may be merged, as depicted in FIG. 5A. In the alternative, different FOVs for different marker images may be selected and merged, as depicted in FIG. 5B.

[0083] Moreover, different FOVs for different marker images may be analyzed independently based on a user's needs. FIGS. 6A-6B depict integrating FOVs, according to an exemplary embodiment of the subject disclosure. With reference to FIG. 6A, all the FOVs are selected and, with reference to FIG. 6B, only the FOVs corresponding to specific markers are selected. Each circle 661 represents a possible FOV for the markers. Each dot 662 in each circle 661 represents a local maximum point for each FOV. Each circle 661 may surround a different marker. Line 663 corresponds to the separation between the tumor and the non-tumor regions. FOVs 664 outside of tumor regions are excluded by morphological operations, such as union and intersection. The final FOVs (i.e., the FOVs that are selected for analysis) are the union of all the FOVs from each marker, as depicted by the methods of FIGS. 5A and 5B.

[0084] In some embodiments, the FOV may be a rectangle about the local maxima. In other embodiments, the FOV may be an arbitrary shape. In some embodiments, the FOV may be a border around a region of high intensity.

[0085] FIG. 6B depicts specifying the most important markers for a given problem by the user, and merging the FOVs based on the selected markers. For example, assume PF3 and CD8 are the most important markers. All the images of single markers may be aligned to the same coordinate system (e.g. the reference coordinate can be the slide section in the middle of the tissue block or the slide with a specific marker) using image registration. Each image may therefore be aligned from its old coordinate system to the new reference coordinate system. FOVs of selected markers (e.g. FP3 and CD8) from an individual marker image may be aligned to the common space and merged using morphological operations such as union and intersection to obtain the merged FOVs (FOVs 665 in FIG. 6B). FIG. 6C shows the morphological operations. Assume A is the FOV from CD8 image and B is the FOV from FP3 image. We first overlay A and B in the same coordinate system and obtain the overlapped region C by computing the intersection of A and B. We then evaluate the ratio of the area of C and the area of A (or B). If the ratio is greater than a threshold (e.g. 0.6, 0.8, etc.), further referred to as an Overlap threshold', we select the FOVs, otherwise we discard the FOVs. The merged FOVs may be mapped back to all the single marker images using inverse registration (i.e. align the registered image in the new coordinate system back to its original old coordinate system) for further analysis. FOVs 664 outside tumor regions are excluded.

[0086] FIGS. 7 and 8 depict user interfaces for image analysis using all marker views and individual markers views, according to exemplary embodiments of the subject disclosure. In these exemplary embodiments, a user interface associated with a computing device may be utilized to perform the FOV selection. The user interface may have All Marker functionalities (FIG. 7) and Single Marker Functionalities (FIG. 8). The marker functions can be accessed by selecting from a tab on the top of the user interface. When using the All Marker function as shown in FIG. 7, all the markers may be viewed and the heat map computation, FOV selection, key marker selection, registration and inverse registration can be performed. In the All Marker View (i.e., a view that illustrates all the markers side by side) options may be provided such as loading a list 771 of image folders(a) with each folder containing all the images including the multiplex and single stains for the same case. Allow batch processing of all the images in the list. Other options provided in a feature panel 772 may include linking the axes for all the images to simultaneously zoom in and out on the images to view the corresponding regions (b), selecting the number of FOVs(c), align the images to a common coordinate system(d), and allowing the user to pick the most important markers for integrating FOVs(e). Colors may be depicted indicating the markers that the FOVs come from. Further options provided may include allowing the user to switch 774 between the heat map view and IHC view, and computing 773 the heat map of each image.

[0087] FIG. 8 depicts the Individual Marker View or Single Marker View, displaying the final selected FOVs for each marker. Features provided in this view may include displaying a thumbnail 881 of the whole slide image, with the FOVs annotated by box in the thumbnail image and a text number near the box indicating the index of the FOV. Other features may include allowing the user to select from the FOV list 883 to delete un-wanted FOVs using checkbox, displaying the high resolution image of the selected FOV 882, saving the image of each FOV into a local folder at original resolution (d), and allowing the user to assign a label to each FOV (e). The labels can be the regions associated with the FOV such as peripheral region, tumor region, and lymphocyte region etc. It will be recognized by those having ordinary skill in the art that these exemplary interfaces may differ from application to application and across various computing technologies, and may use different versions of interface so long as the novel features described herein are enabled in light of this disclosure.

[0088] Therefore, the systems and methods disclosed herein provide automatic FOV selection, and have been found important to analyzing biological specimens, and useful in computing tissue analyses scores, for example in immunoscore computations. Operations disclosed herein overcome disadvantages known in the prior art, such as FOV selection being un- reproducible and biased in human reader manual FOV selection, as the automatic FOV selection is able to provide the FOVs via a computer without relying on a human reader's manual selection. When combined with automatic cell counting, such as immune cell counting, and data analysis, the disclosed operations allow a complete automatic workflow that takes in one or more scanned images or image data as input, and outputs the final clinical outcome prediction. The systems and methods disclosed herein provide automatic FOV selection, and have been found important to analyzing biological specimens, and useful in computing tissue analyses scores, for example in immunoscore computations. Operations disclosed herein overcome disadvantages known in the prior art, such as FOV selection being un-reproducible and biased in human reader manual FOV selection, as the automatic FOV selection is able to provide the FOVs via a computer without relying on a human reader's manual selection. When combined with automatic immune cell counting and data analysis, the disclosed operations allow a complete automatic workflow that takes in one or more scanned images or image data as input, and outputs the final clinical outcome prediction. [0089] FIG. 9 depicts a digital pathology workflow for immunoscore computation, according to an exemplary embodiment of the subject disclosure. This embodiment illustrates how the automatic FOV selection method disclosed herein may be utilized in an immunoscore computation workflow. For example, after a slide is scanned 991 and the FOVs have been selected 992 according to the operations disclosed herein, an automatic detection 993 of different types of cells in each FOV can be performed. The automatic cell detection technique, for example according to the method disclosed in Patent Application US , Serial Number 62/002,633 filed May 23, 2014 and PCT/EP2015/061226, entitled "Deep Learning for Cell Detection", which is hereby incorporated by reference in its entirety, is an exemplary embodiment utilized to obtain detect the cells. Further, features (e.g., features related to the number and/or types of cells identified) can be extracted 994 that are related the one or more cells detected for each biological specimen (e.g., tissue samples, etc.). The features can be number of different types of cells and the ratios of cells in different FOVs related to different regions in the tissue image such as the tumor region and the periphery region. Those features can be used to train 995 a classifier (such as Random Forest and Support Vector Machine) and classify each case to the different outcome classes (e.g., likelihood of relapse or not).

[0090] FIG. 10 depicts a process flow for an exemplary embodiment of the present invention. An input image (1001 ) is received from the image acquisition system. In addition, a series of low-resolution marker images (1004) are received from the image acquisition system. The marker images may be derived by unmixing of the high-resolution image or may be received as single stain slide images. The low resolution input image is used to compute a tissue region mask (1003), which indicates which parts of the image contain tissue of interest. The low resolution image marker images are passed through a low pass filter to produce filtered image marker images (1005). The tissue region mask (cf. tissue mask 233 of Fig. 2) is then applied to the low pass filtered images to block out (reduce to 0) regions that are not of interest. The results in a masked filtered image (1006) for each marker. A local max filter is applied to a max filtered image to identify local maxima (1007). The top K local maxima are selected (1008), and for each local maxima a field of view is defined (1009). Then the FOVs for each image are merged (1010), by transferring all images to a common coordinate frame and overlaying and combining any overlapping fields of view. The merged fields of view are then transferred back to the original image coordinate system, extracting the regions from the high resolution input image for analysis.

[0091] FIG. 1 1 shows a different process flow for another exemplary embodiment of the present invention. The process flow is divided into a FOV generation step (1 100) as shown in FIG. 1 1 a, and a field of view merging step (1 124) as shown in FIG. 1 1 b. In the FOV generation step, single stain images (1 101 ) are received from the image acquisition system. The images are low-pass filtered (1 102). In some embodiments, the images may be converted to a lower resolution (1 103), which speeds processing. In some embodiments an unmixing step (1 104) may be applied to extract the color channel of interest from the single stain slides, if it is not already reduced to a single color channel, producing single marker images (1 108). In some embodiments an HTX image (1 105) may also be generated. The single marker image is then segmented (1 109) to identify features of interest. From the segmented image a tissue region mask (1 1 10) is generated. In some embodiments, the single marker image may be visualized (1 106) using a heat map (1 107), by assigning colors to regions of varying intensity in the single marker image. The tissue region mask (1 1 10) is then applied to the single marker image (1 1 1 1 ), resulting in a foreground image (1 1 12), which displays the intensity of the marker image only in the tissue region of interest. The foreground image is passed through a local max filter (1 1 13), to identify peaks in intensity. Candidate FOV coordinates are identified as the top K peaks of the local max filtered image (1 1 14). Finally, regions around each candidate FOV coordinate are defined (1 1 15) to obtain the list of candidate FOVs (1 1 16). These operations are performed for each single stain slide.

[0092] In the FOV merging step (1 124), all of the candidate FOV lists for the various single stain slides are obtained (1 1 17). The images are registered to a single coordinate frame (1 1 18), by selecting one image as a reference image and transforming the other images to match the reference image. The candidate FOV coordinates are then transformed accordingly to obtain aligned candidate FOV lists (1 1 19). The FOVs are then overlaid and merged (1 120), to obtain a unified FOV list for all images (1 121 ). Inverse registration is then performed (1 122) to transform the unified FOVs back to each of the original coordinate systems of the original single stain images (1 123). The FOVs can then be displayed on the original single stain slides.

[0093] Figure 12 shows process flow of an alternative embodiment of the present invention, using multiplex slides as inputs (1201 ). In the FOV generation step, multiplex slides (1201 ) are received from the image acquisition system. The images are low-pass filtered (1202). In some embodiments, the images may be converted to a lower resolution (1203), which speeds processing. In this embodiment, an unmixing step (1204) is applied to extract the color channels of interest from the multiplex slide, producing a plurality of single marker images (1208). In some embodiments an HTX image (1205) may also be generated. The first single marker image is then segmented (1209) to identify features of interest. From the segmented image a tissue region mask (1210) is generated. In some embodiments, the single marker image may be visualized (1265) using a heat map (1207), by assigning colors to regions of varying intensity in the single marker image. The tissue region mask (1210) is then applied to the single marker image (1210), resulting in a foreground image (1212) which displays the intensity of the marker image only in the tissue region of interest. The foreground image is passed through a local max filter (1213), to identify peaks in intensity. Candidate FOV coordinates are identified as the top K peaks of the local max filtered image

(1214) . Finally, regions around each candidate FOV coordinate are defined

(1215) to obtain the list of candidate FOVs (1216). These operations are performed for each single stain slide in order. The FOV merging step proceeds as in FIG. 1 1 b.

[0094] Figure 13 shows yet another process flow of an alternative embodiment of the present invention, using single stain images (1301 ) as inputs. The images are low-pass filtered (1302). In some embodiments, the images may be converted to a lower resolution (1303), which speeds processing. In some embodiments an unmixing step (1304) may be applied to extract the color channel of interest from the single stain slides, if it is not already reduced to a single color channel, producing single marker images (1308). In some embodiments an HTX image (1305) may also be generated. In other embodiments, the single marker image may be visualized (1306) using a heat map (1307), by assigning colors to regions of varying intensity in the single marker image. In one embodiment, the lower resolution images are segmented (1309) to identify features of interest. From the segmented image, a tissue region mask (1310) is generated and then the mask operation is applied (131 1 ) to the segmented image, resulting in a foreground image (1312), which displays the intensity of the marker image only in the tissue region of interest. In another embodiment, the mask operation (131 1 ) is applied to the single marker image (1308), resulting in a foreground image (1312). In either embodiment, the foreground image (1312) is passed through a local max filter (1313) to identify peaks in intensity. Candidate FOV coordinates are identified as the top K peaks of the local max filtered image (1314). Finally, regions around each candidate FOV coordinate are defined (1315) to obtain the list of candidate FOVs (1316). These operations are performed for each single stain slide. The FOV merging step proceeds as in FIG. 1 1 b.

[0095] The computer-implemented method for automatic FOV selection, in accordance with the present invention, has been described, for exemplary purposes, in connection with the identification of immune cells, and for use in immunoscore computations. However, the computer-implemented method for automatic FOV selection, in accordance with the present invention, is applicable to images of any type of image of a cell or image of a biological specimen, and is applicable to determinations of type, density and location for any type of cell or group of cells. Moreover, besides medical applications such as anatomical or clinical pathology, prostrate / lung cancer diagnosis, etc., the same methods may be performed to analysis other types of samples such as remote sensing of geologic or astronomical data, etc. The operations disclosed herein may be ported into a hardware graphics processing unit (GPU), enabling a multithreaded parallel implementation.

[0096] Computers typically include known components, such as a processor, an operating system, system memory, memory storage devices, input-output controllers, input-output devices, and display devices. It will also be understood by those of ordinary skill in the relevant art that there are many possible configurations and components of a computer and may also include cache memory, a data backup unit, and many other devices. Examples of input devices include a keyboard, a cursor control devices (e.g., a mouse), a microphone, a scanner, and so forth. Examples of output devices include a display device (e.g., a monitor or projector), speakers, a printer, a network card, and so forth. Display devices may include display devices that provide visual information, this information typically may be logically and/or physically organized as an array of pixels. An interface controller may also be included that may comprise any of a variety of known or future software programs for providing input and output interfaces. For example, interfaces may include what are generally referred to as "Graphical User Interfaces" (often referred to as GUI's) that provide one or more graphical representations to a user. Interfaces are typically enabled to accept user inputs using means of selection or input known to those of ordinary skill in the related art. The interface may also be a touch screen device. In the same or alternative embodiments, applications on a computer may employ an interface that includes what are referred to as "command line interfaces" (often referred to as CLI's). CLI's typically provide a text based interaction between an application and a user. Typically, command line interfaces present output and receive input as lines of text through display devices. For example, some implementations may include what are referred to as a "shell" such as Unix Shells known to those of ordinary skill in the related art, or Microsoft Windows Powershell that employs object-oriented type programming architectures such as the Microsoft .NET framework. [0097] Those of ordinary skill in the related art will appreciate that interfaces may include one or more GUI's, CLI's or a combination thereof. A processor may include a commercially available processor such as a Celeron, Core, or Pentium processor made by Intel Corporation, a SPARC processor made by Sun Microsystems, an Athlon, Sempron, Phenom, or Opteron processor made by AMD Corporation, or it may be one of other processors that are or will become available. Some embodiments of a processor may include what is referred to as multi-core processor and/or be enabled to employ parallel processing technology in a single or multi-core configuration. For example, a multi-core architecture typically comprises two or more processor "execution cores". In the present example, each execution core may perform as an independent processor that enables parallel execution of multiple threads. In addition, those of ordinary skill in the related will appreciate that a processor may be configured in what is generally referred to as 32 or 64 bit architectures, or other architectural configurations now known or that may be developed in the future.

[0098] A processor typically executes an operating system, which may be, for example, a Windows type operating system from the Microsoft Corporation; the Mac OS X operating system from Apple Computer Corp.; a Unix or Linux-type operating system available from many vendors or what is referred to as an open source; another or a future operating system; or some combination thereof. An operating system interfaces with firmware and hardware in a well-known manner, and facilitates the processor in coordinating and executing the functions of various computer programs that may be written in a variety of programming languages. An operating system, typically in cooperation with a processor, coordinates and executes functions of the other components of a computer. An operating system also provides scheduling, input-output control, file and data management, memory management, and communication control and related services, all in accordance with known techniques.

[0099] System memory may include any of a variety of known or future memory storage devices that can be used to store the desired information and that can be accessed by a computer. Computer readable storage media may include volatile and non-volatile, removable and non-removable media implemented in any method or technology for storage of information such as computer readable instructions, data structures, program modules, or other data. Examples include any commonly available random access memory (RAM), read-only memory (ROM), electronically erasable programmable read-only memory (EEPROM), digital versatile disks (DVD), magnetic medium, such as a resident hard disk or tape, an optical medium such as a read and write compact disc, or other memory storage device. Memory storage devices may include any of a variety of known or future devices, including a compact disk drive, a tape drive, a removable hard disk drive, USB or flash drive, or a diskette drive. Such types of memory storage devices typically read from, and/or write to, a program storage medium such as, respectively, a compact disk, magnetic tape, removable hard disk, USB or flash drive, or floppy diskette. Any of these program storage media, or others now in use or that may later be developed, may be considered a computer program product. As will be appreciated, these program storage media typically store a computer software program and/or data. Computer software programs, also called computer control logic, typically are stored in system memory and/or the program storage device used in conjunction with memory storage device. In some embodiments, a computer program product is described comprising a computer usable medium having control logic (computer software program, including program code) stored therein. The control logic, when executed by a processor, causes the processor to perform functions described herein. In other embodiments, some functions are implemented primarily in hardware using, for example, a hardware state machine. Implementation of the hardware state machine so as to perform the functions described herein will be apparent to those skilled in the relevant arts. Input-output controllers could include any of a variety of known devices for accepting and processing information from a user, whether a human or a machine, whether local or remote. Such devices include, for example, modem cards, wireless cards, network interface cards, sound cards, or other types of controllers for any of a variety of known input devices. Output controllers could include controllers for any of a variety of known display devices for presenting information to a user, whether a human or a machine, whether local or remote. In the presently described embodiment, the functional elements of a computer communicate with each other via a system bus. Some embodiments of a computer may communicate with some functional elements using network or other types of remote communications. As will be evident to those skilled in the relevant art, an instrument control and/or a data processing application, if implemented in software, may be loaded into and executed from system memory and/or a memory storage device. All or portions of the instrument control and/or data processing applications may also reside in a readonly memory or similar device of the memory storage device, such devices not requiring that the instrument control and/or data processing applications first be loaded through input-output controllers. It will be understood by those skilled in the relevant art that the instrument control and/or data processing applications, or portions of it, may be loaded by a processor, in a known manner into system memory, or cache memory, or both, as advantageous for execution. Also, a computer may include one or more library files, experiment data files, and an internet client stored in system memory. For example, experiment data could include data related to one or more experiments or assays, such as detected signal values, or other values associated with one or more sequencing by synthesis (SBS) experiments or processes. Additionally, an internet client may include an application enabled to access a remote service on another computer using a network and may for instance comprise what are generally referred to as "Web Browsers". In the present example, some commonly employed web browsers include Microsoft Internet Explorer available from Microsoft Corporation, Mozilla Firefox from the Mozilla Corporation, Safari from Apple Computer Corp., Google Chrome from the Google Corporation, or other type of web browser currently known in the art or to be developed in the future. Also, in the same or other embodiments an Internet client may include, or could be an element of, specialized software applications enabled to access remote information via a network such as a data processing application for biological applications.

[00100] A network may include one or more of the many various types of networks well known to those of ordinary skill in the art. For example, a network may include a local or wide area network that may employ what is commonly referred to as a TCP/IP protocol suite to communicate. A network may include a network comprising a worldwide system of interconnected computer networks that is commonly referred to as the Internet, or could also include various intranet architectures. Those of ordinary skill in the related arts will also appreciate that some users in networked environments may prefer to employ what are generally referred to as "firewalls" (also sometimes referred to as Packet Filters, or Border Protection Devices) to control information traffic to and from hardware and/or software systems. For example, firewalls may comprise hardware or software elements or some combination thereof and are typically designed to enforce security policies put in place by users, such as for instance network administrators, etc.

[00101] The foregoing disclosure of the exemplary embodiments of the present subject disclosure has been presented for purposes of illustration and description. It is not intended to be exhaustive or to limit the subject disclosure to the precise forms disclosed. Many variations and modifications of the embodiments described herein will be apparent to one of ordinary skill in the art in light of the above disclosure. The scope of the subject disclosure is to be defined only by the claims appended hereto, and by their equivalents.

[00102] Further, in describing representative embodiments of the present subject disclosure, the specification may have presented the method and/or process of the present subject disclosure as a particular sequence of steps. However, to the extent that the method or process does not rely on the particular order of steps set forth herein, the method or process should not be limited to the particular sequence of steps described. As one of ordinary skill in the art would appreciate, other sequences of steps may be possible. Therefore, the particular order of the steps set forth in the specification should not be construed as limitations on the claims. In addition, the claims directed to the method and/or process of the present subject disclosure should not be limited to the performance of their steps in the order written, and one skilled in the art can readily appreciate that the sequences may be varied and still remain within the spirit and scope of the present subject disclosure.

Claims

WHAT IS CLAIMED IS:

1. An image processing method for analyzing a multi-channel image (231 ) obtained from a biological tissue sample being stained by multiple stains, the method comprising:

a. unmixing the multi-channel image to provide an unmixed image (232, 238) per channel,

b. spatial low pass filtering (234) of at least one (232) of the unmixed images,

c. local maximum filtering (236)of the at least one (234; 235) of the spatial low pass filtered unmixed images,

d. thresholding (237) the at least one of the spatial low pass filtered unmixed images (236) s to identify at least one set of neighboring pixels (239.1 ; 239.4),

e. defining a field of view (240.1 , 240.2; A, B) by extracting an image portion of the multi-channel image (231 ) from an image location given by the set of neighboring pixels (239.1 ; 239.4), the field of view having a predetermined size and shape,

f. analyzing the field of view for determining at least one biological parameter.

2. The method of claim 1 , further comprising:

- segmentation of another one of the unmixed images (238) for identification of tissue regions to provide a tissue region mask (233),

- masking the multi-channel image or the at least one of the unmixed images before or after execution of steps c, d or e with the tissue mask.

3. The method of claim 2, wherein the another one of the unmixed images is representative of one of the stains that is a counter stain to the stain represented by the one of the images.

4. The method of claims 1 , 2 or 3, wherein the steps b to e are executed for at least an additional one of the unmixed images resulting in the definition of an additional field of view and further comprising merging the field of view and the additional field of view before execution of step f, if a degree of spatial overlap of the field of view and the additional field of view is above an overlap threshold.

5. The method of any one of the preceding claims, wherein the analysis is performed by cell counting in order to determine the number of cells within the field of view.

6. The method of any one of the preceding claims, wherein the analysis of the field of view is performed by means of a convolutional neural network (CNN) for determining of a probability for the presence of a biological feature in the field of view.

7. The method of any one of the preceding claims, wherein a data analysis is performed on the image data contained in the field of view for determining the at least one biological parameter, wherein the data analysis is selected from the group consisting of cluster analysis or statistical analysis.

8. The method of any one of the preceding claims, further

- comprising staining the biological tissue sample with the multiple stains to provide the channels,

- acquiring the multi-channel image using an image sensor (102).

9. An image processing system for analyzing a multi-channel image obtained from a biological tissue sample being stained by multiple stains, the image processing system comprising a processing component (105) being configured for execution of: a) unmixing the multi-channel image to provide an unmixed image (232, 238) per channel,

b) spatial low pass filtering (234) of at least one (232) of the unmixed images,

c) local maximum filtering (236)of the at least one (234; 235) of the spatial low pass filtered unmixed images,

d) thresholding (237) the at least one of the spatial low pass filtered unmixed images (236) s to identify at least one set of neighboring pixels (239.1 ; 239.4),

e) defining a field of view (240.1 , 240.2; A, B) by extracting an image portion of the multi-channel image (231 ) from an image location given by the set of neighboring pixels (239.1 ; 239.4), the field of view having a predetermined size and shape,

f) analyzing the field of view for determining at least one biological parameter.

10. The image processing system of claim 9, further comprising a user interface for displaying a field of view list (883) for a user's selection or deselection of one of the field of views for performing the analysis in step f.

1 1 . A method for automatic detection of structures in an image of a biological tissue sample, the method being implemented by an imaging system, the method being stored on a medium readable by a computing device and comprising logical instructions that are executed by a processor to perform operations comprising:

a. receiving a color image;

b. applying a local max filter to the filtered color channel image, wherein the local max filter kernel has a spatial extent on the order of a spatial extent of features enhanced by a spatial filtering applied to the single-color image; c. thresholding the local max filter image to produce a binary image that contains a region of interest; and

d. identifying an isolated binary region as a candidate field of view (FOV).

12. A method for automatic detection of structures in an image of a biological tissue sample, the method being implemented by an imaging system configured to perform operations comprising:

a. receiving a color image;

b. applying a color unmixing operation to the image to produce a single- color image in a channel of interest;

c. applying a spatial frequency filter to the single-color image, wherein the spatial frequencies corresponding to structural features of interest are enhanced, and all other spatial frequencies are diminished; d. applying a local max filter to the filtered color channel image, wherein the local max filter kernel has spatial extent on the order of the spatial extent of features enhanced by the spatial filtering applied to the single-color image;

e. thresholding the local max filter image to produce a binary image that contains a region of interest; and

f. identifying an isolated binary region as a candidate field of view (FOV).

13. The method of claim 12 wherein the binary image identifies a plurality of regional maxima in the image, and then further identifies a plurality of candidate FOVs based on the identified regional maxima.

14. The method of claim 13, further comprising ranking the plurality of candidate FOVs.

15. The method of claim 13, wherein the plurality of candidate of FOVs are ranked in at least one of ascending order or descending order.

16. The method of claim 13, wherein the plurality of candidate FOVs are ranked according to intensity.

17. The method of claim 13, wherein the plurality of candidate FOVs are associated with an intensity value relative to the other intensity values in the image.

18. The method of claim 17, further comprising outputting a predetermined number of the plurality of candidate FOVs.

19. The method of claim 17, further comprising outputting a predetermined number of the plurality of candidate FOVs to a user interface.

20. The method of claim 16 further comprising

a. receiving a predetermined number of the plurality of candidate FOVs to output;

b. ranking the plurality of candidate FOVs in descending order, wherein top highest intensity regions are ranked above lower intensity regions; and

c. selecting the predetermined number of FOVs from the top highest intensity regions.

21 . The method of claim 12, wherein the color image received is a multiplex image.

22. The method of claim 12, wherein the color image is an image of the biological tissue sample stained with at least one IHC stain and hematoxylin.

23. The method of claim 12, wherein the color image is an image of the biological tissue sample stained with an IHC multiplex assay and hematoxylin, wherein the IHC multiplex assay is a combination of stains, and each stain targets a biomarker.

24. The method of claim 22, wherein the color image is unmixed into individual marker channels corresponding to each of the IHC stains in the assay and an individual hematoxylin channel image, and wherein the local max filter is applied only to the individual marker channels.

25. The method of claim 22 further comprising performing a connected components analysis on the identified plurality of candidate FOVs, wherein the connected components analysis identifies a number of the plurality of candidate FOVs.

26. The method of claim 22 further comprising performing a connected components analysis on the identified plurality of candidate FOVs, wherein the connected components analysis labels each of the plurality of candidate FOVs, such that a number of total plurality of candidate FOVs is determined.

27. The method of claim 26 further comprising replacing the assigned intensity value associated with each of the plurality of candidate FOVs by the local max filter with an actual intensity value associated with each of the plurality of candidate FOVs in the color image.

28. The method of claim 27 further comprising sorting the plurality of candidate FOVs based on intensity.

29. The method of claim 28 further comprising selecting a predetermined number of the plurality of candidate FOVs, wherein predetermined number of the plurality of candidate FOVs that are selected are those plurality of candidate FOVs having the highest intensity values

30. The method of claim 29 further comprising finding the center of each of the selected plurality of candidate FOVs.

31 . The method of claim 30, wherein the center of each of the plurality of selected candidate FOVs is determined by computing the mean of the coordinates of each of the pixels in the plurality of candidate FOVs that was determined by the local max filter.

32. The method of claim 30 further comprising segmenting the individual color channel images, wherein the segmentation is performed by forming a neighborhood of pixels around each center of the selected predetermined number of the plurality of candidate FOVs.

33. A system for quality control of automated detection of structures in an image of a biological tissue sample, the system comprising:

a. a processor; and

b. a memory coupled to the processor, the memory stores computer- readable instructions that, when executed by the processor, cause the processor to perform operations comprising:

i. receiving a color image from the image acquisition system;

ii. applying a color unmixing operation to the image to produce a single-color image in a blended color channel of interest;

iii. applying a spatial frequency filter to the single-color image, wherein the spatial frequencies corresponding to structural features of interest are enhanced, and all other spatial frequencies are diminished;

iv. applying a local max filter to the filtered color channel image, wherein the local max filter kernel has spatial extent on the order of the spatial extent of features enhanced by the spatial filtering applied to the single-color image;

v. thresholding the local max filter image to produce a binary image that contains a region of interest; and

vi. identifying an isolated binary region as a candidate field of view (FOV).

34. The system of claim 33, wherein the binary image identifies a plurality of regional maxima in the image, and then further identifies a plurality of candidate FOVs based on the identified regional maxima.

35. The system of claim 33, wherein the color image received is a multiplex image.

36. The system of claim 33, wherein the color image is an image of a biological tissue sample stained with at least one IHC stain and hematoxylin.

37. The system of claim 33, wherein the color image is an image of a biological tissue sample stained with an IHC multiplex assay and hematoxylin, wherein the IHC multiplex assay is a combination of stains, and each stain targets a biomarker.

38. The system of claim 33, wherein the operations further comprises performing a connected components analysis on the identified plurality of candidate FOVs, wherein the connected components analysis identifies a number of the plurality of candidate FOVs.

39. The system of claim 33, wherein the operations further comprises performing a connected components analysis on the identified plurality of candidate FOVs, wherein the connected components analysis labels each of the plurality of candidate FOVs, such that a number of total plurality of candidate FOVs is determined.

40. A tangible, non-transitory computer-readable medium having computer- executable instructions for execution by a processing system, the computer- executable instructions for automatic detection of structures in an image of a biological tissue sample, the computer-executable instructions, when executed, cause the processing system to perform operations comprising: a. receiving a color image from the image acquisition system; b. applying a color unmixing operation to the image to produce a single- color image in a blended color channel of interest; c. applying a spatial frequency filter to the single-color image, wherein the spatial frequencies corresponding to structural features of interest are enhanced, and all other spatial frequencies are diminished; d. applying a local max filter to the filtered color channel image, wherein the local max filter kernel has spatial extent on the order of the spatial extent of features enhanced by the spatial filtering applied to the single-color image; e. thresholding the local max filter image to produce a binary image that contains a region of interest; and f. identifying an isolated binary region as a candidate field of view (FOV).

41 . The computer-readable medium of claim 40, wherein the binary image identifies a plurality of regional maxima in the image, and then further identifies a plurality of candidate FOVs based on the identified regional maxima.

42. The computer-readable medium of claim 40, wherein the color image received is a multiplex image.

43. The computer-readable medium of claim 40, wherein the color image is an image of a biological tissue sample stained with at least one IHC stain and hematoxylin.

44. The computer-readable medium of claim 40, wherein the color image is an image of a biological tissue sample stained with an IHC multiplex assay and hematoxylin, wherein the IHC multiplex assay is a combination of stains, and each stain targets a biomarker.

45. The computer-readable medium of claim 40, wherein the operations further comprises performing a connected components analysis on the identified plurality of candidate FOVs, wherein the connected components analysis identifies a number of the plurality of candidate FOVs.

46. The computer-readable medium of claim 40, wherein the operations further comprises performing a connected components analysis on the identified plurality of candidate FOVs, wherein the connected components analysis labels each of the plurality of candidate FOVs, such that a number of total plurality of candidate FOVs is determined.

47. A system for quality control of automated whole-slide analysis, the system comprising:

a. a processor; and

i. reading a plurality of single stain slides from the image acquisition system,

ii. applying a low pass filter to produce a plurality of single marker images,

iii. performing image segmentation on the filtered single marker images,

iv. obtaining tissue region masks from the segmented single marker images,

v. generating a foreground image for each image, where the foreground image is the tissue region mask multiplied by the single marker image,

vi. identifying a plurality of candidate fields of view (FOVs) from the foreground image,

vii. merging a subset of a plurality of candidate FOVs for each single marker image into a plurality of merged FOVs, and viii. depicting the merged portion of the plurality of candidate FOVs on the single market images.

48. The system of claim 47, wherein the operations further comprise converting the plurality of single marker images to a lower resolution before performing image segmentation.

49. The system of claim 47, wherein an unmixing step is applied to the filtered single marker images to reduce them to the color channel of interest, where the unmixing module uses a reference color matrix to determine what colors correspond to the individual color channels.

50. The system of claim 47, wherein the generation of the tissue region mask comprises:

a. computing the luminance of the image, producing a luminance image, b. applying a standard deviation filter to the luminance image, producing a filtered luminance image, and

c. applying a threshold to filtered luminance image, such that pixels with a luminance above a given threshold are set to one, and pixels below the threshold are set to zero, producing the tissue region mask.

51 . The system of claim 47, wherein the generation of the tissue region mask comprises obtaining the tissue region from image segmentation algorithms.

52. The system of claim 47, wherein the generation of the tissue region mask comprises obtaining the tissue region from manual delineation.

53. The system of claim 50, wherein the luminance image is computed by the formula, Luminance = w_r R + w_g G + w_b B, for each pixel, where R is the red color component of the pixel, G is the green color component of the pixel, and B is the blue color component of the pixel, where w_r is the weight for the red color component, w_g is the weight of the green color component, and w_b is the weight of the blue color component.

54. There system of claim 53, wherein w_r = 0.2989, w_g=0.5870, and w_b=0.1 140.

55. The system of claim 47, wherein the system further comprises generating a heat map by assigning colors to the foreground image, wherein the colors represent the relative intensity of foreground image, and displaying the heat map on a computer screen.

56. The system of claim 47, wherein identifying the candidate fields of view comprises:

a. applying a local max filter to the foreground image to obtain a plurality of local maxima in the foreground image,

b. selecting the top at least two local maxima from the local max filtered image,

c. defining a region around each selected local maxima, where the region is a candidate field of view, and

d. adding the candidate field of view to the plurality of candidate fields of view.

57. The system of claim 56, wherein the region around the local maxima is a rectangle with a size specified by the user.

58. The system of claim 56, wherein the region around the local maxima is any simple closed curve that contains the local maxima.

59. The system of claim 56, wherein the region around the local maxima is defined by an intensity threshold applied to the pixels around the local maxima, such that the region has a border where the region of high intensity drops below the threshold.

60. The system of claim 56, wherein the region around the local maxima is obtained from image segmentation algorithms.

61 . The system of claim 47, wherein merging the candidate fields of view comprises:

a. registering the single stain images to a common coordinate reference frame, comprising:

i. selecting one single stain image to serve as a reference image; and

ii. computing a transformation of each single stain image to the coordinate frame of the reference image;

b. overlaying the candidate FOVs of each single stain image onto the common coordinate reference frame; and

c. combining any overlapping candidate FOVs into a single FOV, wherein the merged FOV is a region of the image that encompasses both of the candidate fields of view.

62. The system of claim 61 , wherein merging the candidate fields of view further comprises overlaying tumor morphology information onto the FOVs and discarding any FOVs not in the tumor region, where the tumor morphology information is obtained from the user.

63. The system of claim 61 , wherein merging the candidate fields of view further comprises discarding any FOVs which have no overlap with any other FOV.

64. The system of claim 61 , wherein merging the plurality of FOVs further comprises:

a. reading a user input selecting the subset of the plurality of candidate FOVs for each single marker image to use, and

b. merging only the selected FOVs for each single marker image.

65. The system of claim 47, wherein depicting the merged fields of view comprises:

a. performing an inverse registration mapping the common coordinate reference frame back to the original coordinate frame of each single stain image, and

b. transferring the merged fields of view back to each of the plurality of single stain images using the inverse registration to depict the merged fields of view in the coordinate frame of the original single stain image.

66. The system of claim 47, wherein a input image can be preprocessed by applying a low pass filter to generate a low resolution image for fast processing.

67. A system for quality control of automated whole-slide analysis, the system comprising:

a. a processor; and

i. reading a multiplex slide from the image acquisition system, wherein the multiplex slide comprises multiple image marker image in separate color channels,

ii. unmixing the multiplex slide to generate each of the plurality of image marker images, where the unmixing module uses a reference color matrix to determine what colors correspond to the individual color channels,

iii. performing image segmentation on the single marker images, iv. obtaining tissue region masks from the segmented single marker images,

v. generating a foreground image for each image , where the foreground image is the tissue region mask multiplied by the single marker image,

vii. selecting a subset of the candidate FOVs of each foreground image,

viii. merging the subset of a plurality of candidate FOVs for all image marker images into a plurality of merged FOVs, and ix. depicting the merged portion of the plurality of candidate FOVs on the input image.

68. A tangible non-transitory computer-readable medium to store computer- readable code that is executed by a processor to perform operations comprising:

a. reading a plurality of single marker images from the image acquisition system, wherein each single marker image is of a single color channel, b. performing image segmentation on the single marker images, c. obtaining tissue region masks from the segmented single marker images,

d. generating a foreground image for each image , where the foreground image is the tissue region mask multiplied by the single marker image, e. identifying a plurality of candidate fields of view (FOVs) from the foreground image,

f. merging a subset of a plurality of candidate FOVs for each single marker image into a plurality of merged FOVs, and

g. depicting the merged portion of the plurality of candidate FOVs on the single market images.

69. A tangible non-transitory computer-readable medium to store computer- readable code that is executed by a processor to perform operations comprising:

a. loading a list of image folders, wherein each folder contains a plurality of images corresponding to a plurality of biomarkers associated with a patient;

b. displaying heat maps for each of the biomarkers identified in the images; and

c. selecting a number of FOVs from at least one of the images or heat maps according to user input.

70. The computer-readable medium of claim 69, wherein the operations further comprise accepting user input to individually or simultaneously zoom in and out on the heat maps or original low resolution images to view one or more regions or corresponding regions.

71 . The computer-readable medium of claim 69, wherein the operations further comprise accepting user input to select one or more of the markers for integrating the selected FOVs.

72. The computer-readable medium of claim 71 , wherein the operations further comprise accepting user input to navigate through each of the selected FOVs, for each individual marker, to at least one of view the images at high power resolution or high power magnification.

73. The computer-readable medium of claim 71 , wherein the operations further comprise accepting user input to add a field of view manually, delete a field of view, and assign a label to a field of view.