A Practical Guide to Modern Digital Pathology
A Practical Guide to Modern Digital Pathology
A Practical Guide to Modern Digital Pathology
A Practical Guide to Modern Digital Pathology
A practical guide to Digital Staining, AI, foundational models, and the future of pathology workflows.
A practical guide to Digital Staining, AI, foundational models, and the future of pathology workflows.
A practical guide to Digital Staining, AI, foundational models, and the future of pathology workflows.
Author
PathScience
PathScience
Published
Jun 17, 2026
Jun 17, 2026
Read Time
20 min
20 min
Why digital pathology terminology is confusing
Digital pathology has evolved faster than the language used to describe it.
What began as a way to scan glass slides and review them on a computer now includes whole slide imaging, image management systems, artificial intelligence, computational pathology, pathology foundation models, Digital Staining, and emerging direct-to-digital workflows.
That creates real confusion. Depending on who is speaking, "digital pathology" may refer to a scanner, a viewer, a laboratory workflow, an AI image analysis system, a digital archive, a foundation model, or a broader transformation in tissue diagnostics. These terms are often used together, but they do not describe the same thing. They sit at different layers of the pathology workflow and solve different problems.
This guide is meant to be a practical field guide. It explains the core technologies and concepts shaping modern digital pathology, where each one fits in the workflow, what people often confuse it with, and why the distinctions matter.
A useful way to think about the field is as a stack:
Tissue is prepared. Stains create visual information. Scanners digitize slides. Software manages and routes the images. Pathologists interpret them. AI can analyze the images, generate new digital stain outputs, or support more intelligent workflows.
The most important distinction is this:
Computational pathology usually analyzes pathology images that already exist. Digital Staining creates digital stain images from image data.
PathScience is focused on the second category: building a Generalized Foundational Model and AI Platform for Pathology that can generate digital stain images directly from unstained brightfield scans, beginning with H&E and developing toward broader digital stain and workflow capabilities
Digital pathology has evolved faster than the language used to describe it.
What began as a way to scan glass slides and review them on a computer now includes whole slide imaging, image management systems, artificial intelligence, computational pathology, pathology foundation models, Digital Staining, and emerging direct-to-digital workflows.
That creates real confusion. Depending on who is speaking, "digital pathology" may refer to a scanner, a viewer, a laboratory workflow, an AI image analysis system, a digital archive, a foundation model, or a broader transformation in tissue diagnostics. These terms are often used together, but they do not describe the same thing. They sit at different layers of the pathology workflow and solve different problems.
This guide is meant to be a practical field guide. It explains the core technologies and concepts shaping modern digital pathology, where each one fits in the workflow, what people often confuse it with, and why the distinctions matter.
A useful way to think about the field is as a stack:
Tissue is prepared. Stains create visual information. Scanners digitize slides. Software manages and routes the images. Pathologists interpret them. AI can analyze the images, generate new digital stain outputs, or support more intelligent workflows.
The most important distinction is this:
Computational pathology usually analyzes pathology images that already exist. Digital Staining creates digital stain images from image data.
PathScience is focused on the second category: building a Generalized Foundational Model and AI Platform for Pathology that can generate digital stain images directly from unstained brightfield scans, beginning with H&E and developing toward broader digital stain and workflow capabilities
Reader takeaway
Most confusion in digital pathology comes from mixing technologies across workflow layers.
Whole slide imaging is not the same as digital pathology. Digital pathology is not the same as computational pathology. Computational pathology is not the same as Digital Staining. And a narrow image-analysis model is not the same as a generalized foundational model for pathology.
The distinctions matter because each layer has different users, infrastructure, validation requirements, regulatory considerations, and adoption paths.
Most confusion in digital pathology comes from mixing technologies across workflow layers.
Whole slide imaging is not the same as digital pathology. Digital pathology is not the same as computational pathology. Computational pathology is not the same as Digital Staining. And a narrow image-analysis model is not the same as a generalized foundational model for pathology.
The distinctions matter because each layer has different users, infrastructure, validation requirements, regulatory considerations, and adoption paths.
Pathology technology stack
Modern pathology can be understood as a stack of related but distinct layers:
Modern pathology can be understood as a stack of related but distinct layers:
Tissue and slide preparation
Staining and image creation
Image acquisition
Digital pathology infrastructure
Computational analysis
Foundation models and AI platforms
Digital Staining and direct-to-digital workflows
Tissue is processed, sectioned, mounted, and prepared for review. H&E, IHC, special stains, and emerging Digital Staining workflows create interpretable visual representations of tissue. Brightfield and fluorescence microscopy or scanning systems capture images. Viewers, archives, case management, image sharing, annotations, and workflow systems make digital images usable. AI and image analysis tools evaluate already-created images for detection, quantification, classification, triage, or decision support. Broadly trained models can support multiple downstream pathology tasks and may enable reusable platform capabilities. Digital Staining models generate digital stain images from input image data, including unstained tissue scans.
This framework is not a regulatory classification or a complete laboratory process map. It is a practical way to place different technologies in the correct layer so conversations become clearer.
Tissue is processed, sectioned, mounted, and prepared for review. H&E, IHC, special stains, and emerging Digital Staining workflows create interpretable visual representations of tissue. Brightfield and fluorescence microscopy or scanning systems capture images. Viewers, archives, case management, image sharing, annotations, and workflow systems make digital images usable. AI and image analysis tools evaluate already-created images for detection, quantification, classification, triage, or decision support. Broadly trained models can support multiple downstream pathology tasks and may enable reusable platform capabilities. Digital Staining models generate digital stain images from input image data, including unstained tissue scans.
This framework is not a regulatory classification or a complete laboratory process map. It is a practical way to place different technologies in the correct layer so conversations become clearer.
A note on validation and intended use
Many technologies discussed in this guide exist at different stages of research, development, validation, regulatory review, and clinical adoption. In pathology, intended use matters.
A tool used for education, research, quality control, workflow support, or clinical diagnosis may require different evidence, validation, and regulatory treatment. For example, the College of American Pathologists recommends that laboratories validate whole slide imaging systems for the intended clinical diagnostic use and setting before implementation. [7]
The same principle applies across AI, image analysis, foundation models, and Digital Staining. A technology may be scientifically promising, but clinical deployment still depends on evidence, intended use, workflow integration, quality systems, and regulatory requirements.
This guide is educational. It is meant to clarify terminology without implying that every technology is ready for every clinical use case.
Many technologies discussed in this guide exist at different stages of research, development, validation, regulatory review, and clinical adoption. In pathology, intended use matters.
A tool used for education, research, quality control, workflow support, or clinical diagnosis may require different evidence, validation, and regulatory treatment. For example, the College of American Pathologists recommends that laboratories validate whole slide imaging systems for the intended clinical diagnostic use and setting before implementation. [7]
The same principle applies across AI, image analysis, foundation models, and Digital Staining. A technology may be scientifically promising, but clinical deployment still depends on evidence, intended use, workflow integration, quality systems, and regulatory requirements.
This guide is educational. It is meant to clarify terminology without implying that every technology is ready for every clinical use case.
PART I
The foundation of pathology
Before discussing artificial intelligence, digital pathology, or Digital Staining, it is important to understand the visual and molecular methods that still anchor routine pathology practice.
Before discussing artificial intelligence, digital pathology, or Digital Staining, it is important to understand the visual and molecular methods that still anchor routine pathology practice.
H&E staining
What it is
H&E stands for hematoxylin and eosin. It is the most widely recognized stain in histopathology and the starting point for much of surgical pathology.
Hematoxylin highlights nuclear material in blue-purple tones, while eosin highlights cytoplasm, collagen, connective tissue, and related structures in pink to red tones. The National Cancer Institute describes H&E as a common laboratory method that uses these two dyes to make different parts of the cell easier to see under the microscope. [3]
H&E stands for hematoxylin and eosin. It is the most widely recognized stain in histopathology and the starting point for much of surgical pathology.
Hematoxylin highlights nuclear material in blue-purple tones, while eosin highlights cytoplasm, collagen, connective tissue, and related structures in pink to red tones. The National Cancer Institute describes H&E as a common laboratory method that uses these two dyes to make different parts of the cell easier to see under the microscope. [3]
Why it matters
H&E matters because it provides a high-information, broadly applicable view of tissue morphology. Tissue architecture, cellular detail, inflammation, fibrosis, necrosis, tumor structure, and many disease patterns can be evaluated from an H&E-stained section.
Leica Biosystems describes H&E as the most widely used stain in histology and histopathology laboratories, and emphasizes that it provides a comprehensive view of tissue microanatomy. [4]
For many cases, H&E provides the first diagnostic answer. In other cases, it guides the pathologist toward additional stains, IHC, molecular testing, or deeper review.
H&E matters because it provides a high-information, broadly applicable view of tissue morphology. Tissue architecture, cellular detail, inflammation, fibrosis, necrosis, tumor structure, and many disease patterns can be evaluated from an H&E-stained section.
Leica Biosystems describes H&E as the most widely used stain in histology and histopathology laboratories, and emphasizes that it provides a comprehensive view of tissue microanatomy. [4]
For many cases, H&E provides the first diagnostic answer. In other cases, it guides the pathologist toward additional stains, IHC, molecular testing, or deeper review.
Common misconception
A common misconception is that H&E is simply one stain among many. In practice, H&E functions as pathology's primary visual language.
Pathologists learn disease through H&E. Textbooks use H&E. Tumor boards often begin with H&E. Digital pathology systems, AI models, and Digital Staining workflows often use H&E as a reference point because it is the visual framework pathologists already know.
A common misconception is that H&E is simply one stain among many. In practice, H&E functions as pathology's primary visual language.
Pathologists learn disease through H&E. Textbooks use H&E. Tumor boards often begin with H&E. Digital pathology systems, AI models, and Digital Staining workflows often use H&E as a reference point because it is the visual framework pathologists already know.
Where it fits in the workflow
H&E sits at the image creation layer. It creates the first familiar visual representation of tissue for pathologist review.
H&E sits at the image creation layer. It creates the first familiar visual representation of tissue for pathologist review.
Do not confuse H&E's importance with chemical inevitability
H&E matters because it provides a familiar and information-rich visual framework. The way H&E images are generated may evolve, but the visual language remains central.
H&E matters because it provides a familiar and information-rich visual framework. The way H&E images are generated may evolve, but the visual language remains central.
Immunohistochemistry (IHC)
What it is
Immunohistochemistry, commonly called IHC, is a laboratory method that uses antibodies to detect specific antigens, usually proteins, within tissue samples.
The National Cancer Institute defines IHC as a method that uses antibodies linked to an enzyme or fluorescent dye so that a target antigen can be seen under the microscope. It is used to help diagnose diseases such as cancer and to help distinguish between different cancer types. [5]
Immunohistochemistry, commonly called IHC, is a laboratory method that uses antibodies to detect specific antigens, usually proteins, within tissue samples.
The National Cancer Institute defines IHC as a method that uses antibodies linked to an enzyme or fluorescent dye so that a target antigen can be seen under the microscope. It is used to help diagnose diseases such as cancer and to help distinguish between different cancer types. [5]
Why it matters
If H&E shows morphology, IHC adds molecular specificity.
Pathologists use IHC to confirm diagnoses, classify tumors, identify tumor origin, evaluate prognostic markers, and support treatment decisions. In oncology, IHC is often central to biomarker-driven care, including the assessment of receptors and checkpoint markers that may inform therapy selection.
IHC is also important because it connects traditional morphology to molecular pathology. It helps answer questions that tissue architecture alone may not resolve.
If H&E shows morphology, IHC adds molecular specificity.
Pathologists use IHC to confirm diagnoses, classify tumors, identify tumor origin, evaluate prognostic markers, and support treatment decisions. In oncology, IHC is often central to biomarker-driven care, including the assessment of receptors and checkpoint markers that may inform therapy selection.
IHC is also important because it connects traditional morphology to molecular pathology. It helps answer questions that tissue architecture alone may not resolve.
Common misconception
IHC is sometimes described as a replacement for H&E. In routine practice, the two are complementary.
H&E usually comes first. IHC is ordered selectively when morphology suggests the need for additional characterization or when a diagnostic or treatment question requires protein-level information.
IHC is sometimes described as a replacement for H&E. In routine practice, the two are complementary.
H&E usually comes first. IHC is ordered selectively when morphology suggests the need for additional characterization or when a diagnostic or treatment question requires protein-level information.
Where it fits in the workflow
IHC also sits at the image creation layer, but it creates a more targeted view of tissue by highlighting specific molecular markers rather than general morphology.
IHC also sits at the image creation layer, but it creates a more targeted view of tissue by highlighting specific molecular markers rather than general morphology.
PART II
How pathology became digital
Digital pathology began by turning physical slides into digital images. That shift created new possibilities for remote review, image sharing, archiving, education, and AI analysis.
Digital pathology began by turning physical slides into digital images. That shift created new possibilities for remote review, image sharing, archiving, education, and AI analysis.
Whole slide imaging (WSI)
What it is
Whole slide imaging is the process of scanning an entire glass slide at high resolution to create a digital slide, commonly called a whole slide image.
The Digital Pathology Association defines whole slide imaging as the acquisition process of creating a whole slide image on a slide scanner. [2]
Whole slide imaging is the process of scanning an entire glass slide at high resolution to create a digital slide, commonly called a whole slide image.
The Digital Pathology Association defines whole slide imaging as the acquisition process of creating a whole slide image on a slide scanner. [2]
Why it matters
WSI made modern digital pathology possible. Once a glass slide can be scanned, it can be reviewed on a screen, shared remotely, stored digitally, used for education, incorporated into tumor boards, and analyzed by software.
The clinical significance of WSI was underscored in 2017, when the FDA permitted marketing of the Philips IntelliSite Pathology Solution, the first WSI system for review and interpretation of digital surgical pathology slides prepared from biopsied tissue. [6]
WSI made modern digital pathology possible. Once a glass slide can be scanned, it can be reviewed on a screen, shared remotely, stored digitally, used for education, incorporated into tumor boards, and analyzed by software.
The clinical significance of WSI was underscored in 2017, when the FDA permitted marketing of the Philips IntelliSite Pathology Solution, the first WSI system for review and interpretation of digital surgical pathology slides prepared from biopsied tissue. [6]
Common misconception
The most common misconception is that whole slide imaging and digital pathology are the same thing.
They are not. WSI is the scanning layer. Digital pathology is the broader ecosystem that uses, manages, shares, reviews, and analyzes the images created by WSI.
The most common misconception is that whole slide imaging and digital pathology are the same thing.
They are not. WSI is the scanning layer. Digital pathology is the broader ecosystem that uses, manages, shares, reviews, and analyzes the images created by WSI.
Where it fits in the workflow
WSI sits at the image acquisition layer. It converts a glass slide into image data.
WSI sits at the image acquisition layer. It converts a glass slide into image data.
Do not confuse WSI with digital pathology
WSI creates the digital slide. Digital pathology is the broader environment for using, managing, sharing, reviewing, and analyzing that slide.
WSI creates the digital slide. Digital pathology is the broader environment for using, managing, sharing, reviewing, and analyzing that slide.
Digital pathology
What it is
Digital pathology refers to the acquisition, management, sharing, and interpretation of pathology information in a digital environment.
The Digital Pathology Association defines digital pathology as a dynamic, image-based environment that enables the acquisition, management, and interpretation of pathology information generated from a digitized glass slide. [1]
Digital pathology refers to the acquisition, management, sharing, and interpretation of pathology information in a digital environment.
The Digital Pathology Association defines digital pathology as a dynamic, image-based environment that enables the acquisition, management, and interpretation of pathology information generated from a digitized glass slide. [1]
Why it matters
Digital pathology changes how pathology information moves through a health system or research organization. It enables remote consultation, distributed workloads, digital archives, education, quality review, AI analysis, and more integrated data workflows.
It also changes the operating model of pathology. A slide no longer needs to be physically moved to be viewed. A case can be shared, routed, annotated, and analyzed digitally.
Digital pathology changes how pathology information moves through a health system or research organization. It enables remote consultation, distributed workloads, digital archives, education, quality review, AI analysis, and more integrated data workflows.
It also changes the operating model of pathology. A slide no longer needs to be physically moved to be viewed. A case can be shared, routed, annotated, and analyzed digitally.
Common misconception
Digital pathology is sometimes used as a catchall for every modern pathology technology. That is understandable, but imprecise.
A viewer is not the same as a scanner. A scanner is not the same as an AI model. An AI model that analyzes a stained image is not the same as a foundational model that generates a digital stain image. Digital pathology is the ecosystem, not one product category.
Digital pathology is sometimes used as a catchall for every modern pathology technology. That is understandable, but imprecise.
A viewer is not the same as a scanner. A scanner is not the same as an AI model. An AI model that analyzes a stained image is not the same as a foundational model that generates a digital stain image. Digital pathology is the ecosystem, not one product category.
Where it fits in the workflow
Digital pathology spans multiple workflow layers, especially image acquisition, management, interpretation, and computational analysis.
Digital pathology spans multiple workflow layers, especially image acquisition, management, interpretation, and computational analysis.
Brightfield Microscopy & Scanning
What it is
Brightfield uses transmitted white light to visualize tissue, typically after a stain such as H&E, chromogenic IHC, or a special stain has been applied.
H&E-stained slides are observed using brightfield illumination, and brightfield microscopy remains highly familiar to pathologists. [19]
In digital workflows, brightfield scanners can capture stained slides for digital review. Emerging direct-to-digital workflows can also begin with unstained brightfield scans that become input data for Digital Staining models.
Brightfield uses transmitted white light to visualize tissue, typically after a stain such as H&E, chromogenic IHC, or a special stain has been applied.
H&E-stained slides are observed using brightfield illumination, and brightfield microscopy remains highly familiar to pathologists. [19]
In digital workflows, brightfield scanners can capture stained slides for digital review. Emerging direct-to-digital workflows can also begin with unstained brightfield scans that become input data for Digital Staining models.
Why it matters
Brightfield matters because it is the routine operating environment for much of diagnostic pathology.
Existing lab workflows, scanner infrastructure, image viewers, pathologist training, and quality processes are largely built around brightfield review.
For solid tumor diagnosis, brightfield microscopy using H&E and IHC is widely described as a preferred method for
pathologists. [16]
Brightfield is also important for Digital Staining because it can provide a practical, familiar image acquisition pathway for generating digital stains from unstained tissue scans.
Brightfield matters because it is the routine operating environment for much of diagnostic pathology.
Existing lab workflows, scanner infrastructure, image viewers, pathologist training, and quality processes are largely built around brightfield review.
For solid tumor diagnosis, brightfield microscopy using H&E and IHC is widely described as a preferred method for
pathologists. [16]
Brightfield is also important for Digital Staining because it can provide a practical, familiar image acquisition pathway for generating digital stains from unstained tissue scans.
Common misconception
Brightfield is sometimes treated as less advanced than fluorescence because it is older and more routine. That misses the point.
Brightfield is valuable because it is familiar, operationally integrated, and central to routine pathology. New technologies that fit into brightfield workflows may face different adoption considerations than technologies requiring new imaging ecosystems.
Brightfield is sometimes treated as less advanced than fluorescence because it is older and more routine. That misses the point.
Brightfield is valuable because it is familiar, operationally integrated, and central to routine pathology. New technologies that fit into brightfield workflows may face different adoption considerations than technologies requiring new imaging ecosystems.
Where it fits in the workflow
Brightfield sits at the image acquisition layer and is closely tied to the image creation layer because it is used to view H&E, chromogenic IHC, special stains, and, in emerging workflows, unstained tissue scans that can serve as inputs for Digital Staining.
Brightfield sits at the image acquisition layer and is closely tied to the image creation layer because it is used to view H&E, chromogenic IHC, special stains, and, in emerging workflows, unstained tissue scans that can serve as inputs for Digital Staining.
Fluorescence Microscopy & Scanning
What it is
Fluorescence uses fluorescent labels and specialized imaging systems to visualize specific biological targets. Instead of relying on light absorbed by a stain, fluorescence imaging detects emitted light from labeled structures.
Fluorescence can support immunofluorescence, FISH, multiplex biomarker analysis, spatial biology, and other research and specialized clinical workflows.
Fluorescence uses fluorescent labels and specialized imaging systems to visualize specific biological targets. Instead of relying on light absorbed by a stain, fluorescence imaging detects emitted light from labeled structures.
Fluorescence can support immunofluorescence, FISH, multiplex biomarker analysis, spatial biology, and other research and specialized clinical workflows.
Why it matters
Fluorescence can provide highly specific molecular and spatial information. It is especially valuable when researchers or clinicians need to evaluate multiple markers, spatial relationships, or molecular features within tissue.
Multiplex immunofluorescence can be powerful, but reviews also note practical limitations such as cost, time, specialized equipment, training needs, autofluorescence, and photobleaching considerations. [17]
Fluorescence can provide highly specific molecular and spatial information. It is especially valuable when researchers or clinicians need to evaluate multiple markers, spatial relationships, or molecular features within tissue.
Multiplex immunofluorescence can be powerful, but reviews also note practical limitations such as cost, time, specialized equipment, training needs, autofluorescence, and photobleaching considerations. [17]
Common misconception
Brightfield and fluorescence are often framed as competitors. A more practical framing is that they are different tools for different jobs.
Brightfield is optimized for routine morphology-first workflows. Fluorescence is powerful when specific molecular or multiplex information is needed.
Brightfield and fluorescence are often framed as competitors. A more practical framing is that they are different tools for different jobs.
Brightfield is optimized for routine morphology-first workflows. Fluorescence is powerful when specific molecular or multiplex information is needed.
Where it fits in the workflow
Fluorescence sits at the image acquisition layer and often connects to molecular or multiplex image creation workflows.
Fluorescence sits at the image acquisition layer and often connects to molecular or multiplex image creation workflows.
PART III
How AI changed pathology
Once pathology images became digital, they became computable.
That shift created the foundation for computational pathology, AI-assisted image analysis, computer-aided diagnosis, pathology foundation models, and newer Digital Staining workflows.
But these categories should not be collapsed into one idea.
Some AI systems analyze images that already exist. Others generate new digital stain images from input scan data. Foundation models may support either or both, depending on how they are trained and what they are designed to do.
Once pathology images became digital, they became computable.
That shift created the foundation for computational pathology, AI-assisted image analysis, computer-aided diagnosis, pathology foundation models, and newer Digital Staining workflows.
But these categories should not be collapsed into one idea.
Some AI systems analyze images that already exist. Others generate new digital stain images from input scan data. Foundation models may support either or both, depending on how they are trained and what they are designed to do.
Computational pathology
What it is
Computational pathology applies image analysis, machine learning, artificial intelligence, and data science to pathology images and related data.
A Digital Pathology Association white paper defines terminology and concepts for computational pathology and discusses the role of deep learning, digital image analysis, and clinical implementation challenges. [8]
In practical terms, much of computational pathology operates on images that have already been created for human review, such as H&E whole slide images, IHC images, or other digitized slides.
Computational pathology can help quantify features that are difficult to measure manually, identify patterns across large datasets, support workflow triage, and generate predictions related to diagnosis, prognosis, biomarkers, or quality control.
Common applications include tumor detection, cell segmentation, grading support, biomarker quantification, tissue classification, quality control, and research discovery.
Reviews of artificial intelligence and computational pathology emphasize the potential for AI to improve diagnostic workflows, support patient care, and reduce variability, while also noting implementation challenges. [9]
Computational pathology applies image analysis, machine learning, artificial intelligence, and data science to pathology images and related data.
A Digital Pathology Association white paper defines terminology and concepts for computational pathology and discusses the role of deep learning, digital image analysis, and clinical implementation challenges. [8]
In practical terms, much of computational pathology operates on images that have already been created for human review, such as H&E whole slide images, IHC images, or other digitized slides.
Computational pathology can help quantify features that are difficult to measure manually, identify patterns across large datasets, support workflow triage, and generate predictions related to diagnosis, prognosis, biomarkers, or quality control.
Common applications include tumor detection, cell segmentation, grading support, biomarker quantification, tissue classification, quality control, and research discovery.
Reviews of artificial intelligence and computational pathology emphasize the potential for AI to improve diagnostic workflows, support patient care, and reduce variability, while also noting implementation challenges. [9]
Computational pathology vs. computer-aided diagnosis
Computer-aided diagnosis, or CAD, is best understood as a narrower category within the broader computational pathology landscape.
CAD systems are designed to assist with diagnostic interpretation. Computational pathology is broader. It may include CAD, but it can also include quantification, triage, image search, research discovery, workflow support, quality control, biomarker prediction, and other forms of image-based analysis.
Computer-aided diagnosis, or CAD, is best understood as a narrower category within the broader computational pathology landscape.
CAD systems are designed to assist with diagnostic interpretation. Computational pathology is broader. It may include CAD, but it can also include quantification, triage, image search, research discovery, workflow support, quality control, biomarker prediction, and other forms of image-based analysis.
Common misconception
Computational pathology is sometimes treated as synonymous with digital pathology. The distinction is important.
Digital pathology creates, manages, shares, and operationalizes digital images. Computational pathology analyzes pathology images and related data. It is one major layer within the digital pathology ecosystem.
Another common misconception is that computational pathology and Digital Staining are the same thing. They are not.
Computational pathology usually analyzes images after they have already been created. Digital Staining generates digital stain images from image data.
Computational pathology is sometimes treated as synonymous with digital pathology. The distinction is important.
Digital pathology creates, manages, shares, and operationalizes digital images. Computational pathology analyzes pathology images and related data. It is one major layer within the digital pathology ecosystem.
Another common misconception is that computational pathology and Digital Staining are the same thing. They are not.
Computational pathology usually analyzes images after they have already been created. Digital Staining generates digital stain images from image data.
Where it fits in the workflow
Computational pathology sits primarily at the computational analysis layer. In most workflows, it operates after images have already been created and acquired.
Computational pathology sits primarily at the computational analysis layer. In most workflows, it operates after images have already been created and acquired.
Why this distinction matters
Many computational pathology systems work on human-interpretable images.
Many computational pathology systems work on human-interpretable images.
That is similar to other areas of medical imaging, where rich acquisition data is transformed into images optimized for human review. Those human-readable images are essential, but they are not always the only representation available in the data.
That is similar to other areas of medical imaging, where rich acquisition data is transformed into images optimized for human review. Those human-readable images are essential, but they are not always the only representation available in the data.
Digital Staining changes the question
Instead of asking only, "What can AI detect in this already-created image?" Digital Staining asks, "Can AI transform input scan data into a validated digital stain image that supports interpretation or downstream workflow?"
That is a different technical problem and a different workflow layer.
Instead of asking only, "What can AI detect in this already-created image?" Digital Staining asks, "Can AI transform input scan data into a validated digital stain image that supports interpretation or downstream workflow?"
That is a different technical problem and a different workflow layer.
Foundation models in pathology
What it is
Foundation models are large AI models trained on broad datasets and designed to support many downstream tasks. In pathology, these models are typically trained on large collections of histopathology images, sometimes combined with reports or text.
The key idea is reuse. Instead of building a separate model from scratch for every task, a foundation model can provide a general representation that can be adapted or evaluated across many applications.
Foundation models are large AI models trained on broad datasets and designed to support many downstream tasks. In pathology, these models are typically trained on large collections of histopathology images, sometimes combined with reports or text.
The key idea is reuse. Instead of building a separate model from scratch for every task, a foundation model can provide a general representation that can be adapted or evaluated across many applications.
Why it matters
Foundation models are shifting pathology AI from narrow point solutions toward broader platform capabilities.
Examples from the research literature include UNI, a general-purpose self-supervised model pretrained on more than 100 million images from more than 100,000 diagnostic H&E-stained whole slide images; Prov-GigaPath, an open-weight whole-slide foundation model trained on large-scale real-world pathology data; and TITAN, a multimodal whole-slide model trained with visual self-supervision and vision-language alignment. [10] [11] [12]
For pathologists and researchers, foundation models may support classification, biomarker prediction, tissue phenotyping, retrieval, report-aligned reasoning, and future image-generation workflows. For investors and technology leaders, they represent a possible platform layer rather than a single application.
Foundation models are shifting pathology AI from narrow point solutions toward broader platform capabilities.
Examples from the research literature include UNI, a general-purpose self-supervised model pretrained on more than 100 million images from more than 100,000 diagnostic H&E-stained whole slide images; Prov-GigaPath, an open-weight whole-slide foundation model trained on large-scale real-world pathology data; and TITAN, a multimodal whole-slide model trained with visual self-supervision and vision-language alignment. [10] [11] [12]
For pathologists and researchers, foundation models may support classification, biomarker prediction, tissue phenotyping, retrieval, report-aligned reasoning, and future image-generation workflows. For investors and technology leaders, they represent a possible platform layer rather than a single application.
Common misconception
Foundation models are sometimes described as if they automatically solve clinical deployment. They do not. Scale is useful, but intended-use validation, workflow integration, data quality, generalization, human factors, and regulatory considerations still matter.
Another misconception is that every pathology foundation model is designed for the same purpose.
Some foundation models are primarily representation models for analyzing existing images. Others may support multimodal reasoning, retrieval, reporting, or image generation. A foundational model designed for Digital Staining is different from a model designed only to classify an existing H&E image.
Foundation models are sometimes described as if they automatically solve clinical deployment. They do not. Scale is useful, but intended-use validation, workflow integration, data quality, generalization, human factors, and regulatory considerations still matter.
Another misconception is that every pathology foundation model is designed for the same purpose.
Some foundation models are primarily representation models for analyzing existing images. Others may support multimodal reasoning, retrieval, reporting, or image generation. A foundational model designed for Digital Staining is different from a model designed only to classify an existing H&E image.
Where it fits in the workflow
Foundation models sit at the AI platform layer. Depending on design, they may support computational analysis, image retrieval, multimodal reasoning, workflow intelligence, Digital Staining, or future clinical and research applications.
Foundation models sit at the AI platform layer. Depending on design, they may support computational analysis, image retrieval, multimodal reasoning, workflow intelligence, Digital Staining, or future clinical and research applications.
PART IV
New approaches to image creation
Historically, pathology innovation focused on image acquisition and image analysis. Increasingly, researchers and companies are also exploring the image creation layer, where traditional staining, digital image generation, and AI meet.
This is where Digital Staining becomes important.
Historically, pathology innovation focused on image acquisition and image analysis. Increasingly, researchers and companies are also exploring the image creation layer, where traditional staining, digital image generation, and AI meet.
This is where Digital Staining becomes important.
Unstained tissue imaging
What it is
Unstained tissue imaging is the capture of images from tissue sections before chemical staining has been applied.
To a pathologist, unstained tissue is usually not directly interpretable in the same way as H&E because it lacks the familiar contrast that stains provide. But as AI methods improve, unstained tissue images can become valuable inputs for computational workflows.
Unstained tissue imaging is the capture of images from tissue sections before chemical staining has been applied.
To a pathologist, unstained tissue is usually not directly interpretable in the same way as H&E because it lacks the familiar contrast that stains provide. But as AI methods improve, unstained tissue images can become valuable inputs for computational workflows.
Why it matters
Unstained tissue imaging is important because it can preserve the same tissue section for downstream use and provide input data for Digital Staining or other computational image-generation methods.
A 2023 Laboratory Investigation study evaluated unstained tissue imaging and digital H&E generation of whole slide images, describing whole-slide unstained microscopy as a feasible approach for computational staining while sparing the same tissue section for subsequent methods. [15]
Unstained tissue imaging is important because it can preserve the same tissue section for downstream use and provide input data for Digital Staining or other computational image-generation methods.
A 2023 Laboratory Investigation study evaluated unstained tissue imaging and digital H&E generation of whole slide images, describing whole-slide unstained microscopy as a feasible approach for computational staining while sparing the same tissue section for subsequent methods. [15]
Common misconception
Unstained tissue imaging is sometimes misunderstood as simply looking at tissue without a stain.
In emerging workflows, the unstained image is not necessarily the final image for review. It may be the input for computational image generation.
Unstained tissue imaging is sometimes misunderstood as simply looking at tissue without a stain.
In emerging workflows, the unstained image is not necessarily the final image for review. It may be the input for computational image generation.
Where it fits in the workflow
Unstained tissue imaging sits between image preparation, image acquisition, and computational image creation. It is the input layer for several Digital Staining approaches.
Unstained tissue imaging sits between image preparation, image acquisition, and computational image creation. It is the input layer for several Digital Staining approaches.
10. Digital Staining
What it is
Digital Staining is the use of computational methods, including AI, to generate digital stain images from tissue image data.
A Digital Staining workflow may begin with an unstained tissue scan, a brightfield image, a fluorescence image, or another imaging input, depending on the system. The output is a digital stain image designed for a specific workflow, such as digital H&E generation or future digital IHC, special stain, or biomarker-related applications.
The term matters because "digital" better describes the goal: creating a usable digital stain output, not a lesser substitute for staining.
Digital Staining is the use of computational methods, including AI, to generate digital stain images from tissue image data.
A Digital Staining workflow may begin with an unstained tissue scan, a brightfield image, a fluorescence image, or another imaging input, depending on the system. The output is a digital stain image designed for a specific workflow, such as digital H&E generation or future digital IHC, special stain, or biomarker-related applications.
The term matters because "digital" better describes the goal: creating a usable digital stain output, not a lesser substitute for staining.
Why it matters
Digital Staining changes how the image creation layer can work.
In a traditional workflow, a tissue section is physically stained and then reviewed under a microscope or scanned into a digital slide.
In a Digital Staining workflow, image data can be captured first and transformed computationally into a digital stain image.
This creates a different set of possibilities:
Digital Staining may help preserve tissue. It may reduce dependency on some chemical staining steps for selected use cases. It may allow multiple digital stain outputs from related source data. It may support faster workflows. It may also create a foundation for more intelligent pathology platforms.
The practical value depends heavily on the input modality, model performance, tissue type, intended use, validation, and workflow integration.
Digital Staining changes how the image creation layer can work.
In a traditional workflow, a tissue section is physically stained and then reviewed under a microscope or scanned into a digital slide.
In a Digital Staining workflow, image data can be captured first and transformed computationally into a digital stain image.
This creates a different set of possibilities:
Digital Staining may help preserve tissue. It may reduce dependency on some chemical staining steps for selected use cases. It may allow multiple digital stain outputs from related source data. It may support faster workflows. It may also create a foundation for more intelligent pathology platforms.
The practical value depends heavily on the input modality, model performance, tissue type, intended use, validation, and workflow integration.
Common misconception
The most important misconception is that Digital Staining is one technology. It is not.
Digital Staining is a category. Different approaches may use different inputs, imaging systems, model architectures, training data, outputs, validation strategies, and intended uses.
The right question is not simply, "Does it do Digital Staining?"
The better questions are:
What is the input?
What digital stain is being generated?
What tissue type is involved?
What scanner or imaging method is used?
What is the intended use?
How has performance been validated?
How does it fit into the pathology workflow?
The most important misconception is that Digital Staining is one technology. It is not.
Digital Staining is a category. Different approaches may use different inputs, imaging systems, model architectures, training data, outputs, validation strategies, and intended uses.
The right question is not simply, "Does it do Digital Staining?"
The better questions are:
What is the input?
What digital stain is being generated?
What tissue type is involved?
What scanner or imaging method is used?
What is the intended use?
How has performance been validated?
How does it fit into the pathology workflow?
Where it fits in the workflow
Digital Staining sits at the image creation layer. It uses computation to generate digital stain images rather than relying only on a separate physical staining process for every view of tissue.
Digital Staining sits at the image creation layer. It uses computation to generate digital stain images rather than relying only on a separate physical staining process for every view of tissue.
11. Digital H&E
What it is
Digital H&E is an AI-generated H&E image created from tissue image data, such as an unstained tissue scan. It is one application of Digital Staining.
Digital H&E is an AI-generated H&E image created from tissue image data, such as an unstained tissue scan. It is one application of Digital Staining.
Why it matters
Digital H&E matters because H&E is the visual framework pathologists already know.
If a Digital Staining workflow can produce a familiar H&E image from an unstained scan, it may fit more naturally into pathologist review than a less familiar imaging output.
That does not mean every digital H&E image is automatically interchangeable with a physical H&E slide for every use case.
Morphologic fidelity, validation design, image quality, intended use, workflow context, and regulatory requirements all matter.
Digital H&E matters because H&E is the visual framework pathologists already know.
If a Digital Staining workflow can produce a familiar H&E image from an unstained scan, it may fit more naturally into pathologist review than a less familiar imaging output.
That does not mean every digital H&E image is automatically interchangeable with a physical H&E slide for every use case.
Morphologic fidelity, validation design, image quality, intended use, workflow context, and regulatory requirements all matter.
Common misconception
The phrase "looks like H&E" can be misleading if it is interpreted too casually.
For pathology, the question is not only whether an image has a blue and pink color palette. The important question is whether the generated digital H&E image preserves the morphologic and contextual information needed for the intended application.
The phrase "looks like H&E" can be misleading if it is interpreted too casually.
For pathology, the question is not only whether an image has a blue and pink color palette. The important question is whether the generated digital H&E image preserves the morphologic and contextual information needed for the intended application.
Where it fits in the workflow
Digital H&E sits at the image creation layer as a specific Digital Staining output that uses the familiar H&E visual framework.
Digital H&E sits at the image creation layer as a specific Digital Staining output that uses the familiar H&E visual framework.
12. Direct-to-Digital Staining
What it is
Direct-to-Digital Staining is PathScience's term for generating digital stain images directly from unstained tissue scans using AI-enabled workflows.
PathScience's platform begins with H&E digital image generation from unstained tissue scans and is being developed toward future digital stain and intelligent workflow applications. [18]
Direct-to-Digital Staining is PathScience's term for generating digital stain images directly from unstained tissue scans using AI-enabled workflows.
PathScience's platform begins with H&E digital image generation from unstained tissue scans and is being developed toward future digital stain and intelligent workflow applications. [18]
Why it matters
The term is useful because it describes a workflow distinction that broader phrases like "digital pathology," "AI pathology," or "Digital Staining" do not always capture.
In a conventional stained-then-scanned workflow, tissue is physically stained first and digitized afterward. In a Direct-to-Digital workflow, the tissue is imaged unstained and the stain image is generated digitally.
The practical implications may include tissue preservation, fewer staining-related workflow steps for selected applications, faster image availability, and the ability to create familiar H&E digital images as part of a digital pathology workflow.
Longer term, the platform opportunity may include additional digital stain types, companion-diagnostic-adjacent workflows, and intelligent workflow support, depending on development, validation, regulatory strategy, and intended use.
The term is useful because it describes a workflow distinction that broader phrases like "digital pathology," "AI pathology," or "Digital Staining" do not always capture.
In a conventional stained-then-scanned workflow, tissue is physically stained first and digitized afterward. In a Direct-to-Digital workflow, the tissue is imaged unstained and the stain image is generated digitally.
The practical implications may include tissue preservation, fewer staining-related workflow steps for selected applications, faster image availability, and the ability to create familiar H&E digital images as part of a digital pathology workflow.
Longer term, the platform opportunity may include additional digital stain types, companion-diagnostic-adjacent workflows, and intelligent workflow support, depending on development, validation, regulatory strategy, and intended use.
Common misconception
Direct-to-Digital Staining should not be used as a synonym for all Digital Staining.
Digital Staining is the broader category.
Direct-to-Digital Staining is the workflow-specific term PathScience uses for generating digital stain images directly from unstained tissue scans.
Direct-to-Digital Staining should not be used as a synonym for all Digital Staining.
Digital Staining is the broader category.
Direct-to-Digital Staining is the workflow-specific term PathScience uses for generating digital stain images directly from unstained tissue scans.
Where it fits in the workflow
Direct-to-Digital Staining sits at the image creation layer and depends on image acquisition, AI modeling, and digital pathology workflow integration.
It connects unstained tissue imaging, brightfield scanning, digital H&E generation, foundation models, and modern pathology workflow infrastructure.
Direct-to-Digital Staining sits at the image creation layer and depends on image acquisition, AI modeling, and digital pathology workflow integration.
It connects unstained tissue imaging, brightfield scanning, digital H&E generation, foundation models, and modern pathology workflow infrastructure.
At PathScience, we are focused on Direct-to-Digital Staining: generating H&E digital stain images directly from unstained tissue scans and developing toward additional Digital Staining and intelligent pathology workflow applications.
At PathScience, we are focused on Direct-to-Digital Staining: generating H&E digital stain images directly from unstained tissue scans and developing toward additional Digital Staining and intelligent pathology workflow applications.
PathScience's Generalized Foundational Model and AI Platform for Pathology
What it is
PathScience is building a Generalized Foundational Model and AI Platform for Pathology.
The goal is not simply to analyze already-created pathology images. The goal is to transform unstained tissue image data into digital stain images that pathologists can use within modern digital workflows.
What it is
PathScience is building a Generalized Foundational Model and AI Platform for Pathology.
The goal is not simply to analyze already-created pathology images. The goal is to transform unstained tissue image data into digital stain images that pathologists can use within modern digital workflows.
PathScience's mission can be summarized this way:
To build the Generalized Foundational Model and AI Platform for Pathology - transforming diagnosis through instant Digital Staining, enhanced precision, and accelerated, intelligent clinical workflows.
To build the Generalized Foundational Model and AI Platform for Pathology - transforming diagnosis through instant Digital Staining, enhanced precision, and accelerated, intelligent clinical workflows.
Why "generalized" matters
The word "generalized" is important.
A narrow AI model may be trained for one stain, one tissue type, one scanner, one workflow, or one task. A generalized foundational model is designed to become a broader platform layer.
For PathScience, the long-term opportunity is to support Digital Staining across a large number of stains, scanners, tissue contexts, and workflow applications over time.
That includes direct-to-digital H&E generation today and development toward additional IHC, special stain, companion diagnostic, and workflow intelligence applications as they are validated and disclosed.
Why this is different from conventional computational pathology
Most computational pathology tools analyze human-interpretable images that already exist.
For example, an AI system may analyze an H&E whole slide image to detect tumor, count cells, quantify a biomarker, or support triage. These tools can be valuable, but they generally operate after the image has been created for human review.
PathScience's Digital Staining foundational model operates earlier in the stack.
It uses unstained brightfield scan data as an input and generates a digital stain image as an output. That means the model is participating in image creation, not only image analysis.
Why this matters technically
Human review depends on images that are optimized for human vision.
That is essential for diagnosis, but human vision is also constrained. Humans interpret color, contrast, brightness, morphology, and context through a limited visual channel.
AI models can operate on image data mathematically and learn relationships that are not limited to the same visual perception constraints. In Digital Staining, the model can learn the relationship between unstained tissue image data and stained outputs across pixels, structures, patterns, and tissue contexts.
This does not mean a model should be described as replacing the pathologist.
A better framing is that PathScience is building an AI platform that can create digital stain images from tissue data in ways that support pathologist review, workflow acceleration, and future intelligent applications.
Why "generalized" matters
The word "generalized" is important.
A narrow AI model may be trained for one stain, one tissue type, one scanner, one workflow, or one task. A generalized foundational model is designed to become a broader platform layer.
For PathScience, the long-term opportunity is to support Digital Staining across a large number of stains, scanners, tissue contexts, and workflow applications over time.
That includes direct-to-digital H&E generation today and development toward additional IHC, special stain, companion diagnostic, and workflow intelligence applications as they are validated and disclosed.
Why this is different from conventional computational pathology
Most computational pathology tools analyze human-interpretable images that already exist.
For example, an AI system may analyze an H&E whole slide image to detect tumor, count cells, quantify a biomarker, or support triage. These tools can be valuable, but they generally operate after the image has been created for human review.
PathScience's Digital Staining foundational model operates earlier in the stack.
It uses unstained brightfield scan data as an input and generates a digital stain image as an output. That means the model is participating in image creation, not only image analysis.
Why this matters technically
Human review depends on images that are optimized for human vision.
That is essential for diagnosis, but human vision is also constrained. Humans interpret color, contrast, brightness, morphology, and context through a limited visual channel.
AI models can operate on image data mathematically and learn relationships that are not limited to the same visual perception constraints. In Digital Staining, the model can learn the relationship between unstained tissue image data and stained outputs across pixels, structures, patterns, and tissue contexts.
This does not mean a model should be described as replacing the pathologist.
A better framing is that PathScience is building an AI platform that can create digital stain images from tissue data in ways that support pathologist review, workflow acceleration, and future intelligent applications.
Common misconception
The common misconception is that PathScience is simply another computational pathology company.
That framing is incomplete.
Computational pathology often means AI image analysis on existing slides. PathScience is building a generalized foundational model for Digital Staining and pathology workflow intelligence.
That places PathScience closer to the image creation and AI platform layer of digital pathology, not only the image analysis layer.
The common misconception is that PathScience is simply another computational pathology company.
That framing is incomplete.
Computational pathology often means AI image analysis on existing slides. PathScience is building a generalized foundational model for Digital Staining and pathology workflow intelligence.
That places PathScience closer to the image creation and AI platform layer of digital pathology, not only the image analysis layer.
Where it fits in the workflow
PathScience's generalized foundational model connects several layers of the pathology stack:
Unstained tissue imaging
Brightfield scanning
AI-based digital image creation
Digital H&E generation
Future Digital Staining applications
Workflow intelligence
Pathologist-facing digital pathology systems
This is the architectural shift: PathScience is not only asking what AI can find in a slide. It is asking how AI can help create the digital pathology images and workflows of the future.
PathScience's generalized foundational model connects several layers of the pathology stack:
Unstained tissue imaging
Brightfield scanning
AI-based digital image creation
Digital H&E generation
Future Digital Staining applications
Workflow intelligence
Pathologist-facing digital pathology systems
This is the architectural shift: PathScience is not only asking what AI can find in a slide. It is asking how AI can help create the digital pathology images and workflows of the future.
How to use this terminology in practice
The most useful digital pathology conversations begin by identifying the workflow layer being discussed.
If the conversation is about scanning glass slides, the relevant term is usually whole slide imaging.
If the conversation is about case review, storage, routing, archiving, or image management, digital pathology is the broader term.
If the conversation is about software extracting information from existing images, computational pathology is more precise.
If the conversation is about AI assisting diagnostic interpretation, computer-aided diagnosis may be the more specific term.
If the conversation is about a broadly trained AI model that can support many downstream pathology tasks, the relevant concept is a pathology foundation model.
If the conversation is about generating digital stain images from tissue image data, the correct category is Digital Staining.
If the workflow begins with unstained tissue scans and generates digital stain images directly, Direct-to-Digital Staining is the more specific PathScience terminology.
This distinction matters because each layer has different users, infrastructure, evidence requirements, operational constraints, and adoption paths.
The most useful digital pathology conversations begin by identifying the workflow layer being discussed.
If the conversation is about scanning glass slides, the relevant term is usually whole slide imaging.
If the conversation is about case review, storage, routing, archiving, or image management, digital pathology is the broader term.
If the conversation is about software extracting information from existing images, computational pathology is more precise.
If the conversation is about AI assisting diagnostic interpretation, computer-aided diagnosis may be the more specific term.
If the conversation is about a broadly trained AI model that can support many downstream pathology tasks, the relevant concept is a pathology foundation model.
If the conversation is about generating digital stain images from tissue image data, the correct category is Digital Staining.
If the workflow begins with unstained tissue scans and generates digital stain images directly, Direct-to-Digital Staining is the more specific PathScience terminology.
This distinction matters because each layer has different users, infrastructure, evidence requirements, operational constraints, and adoption paths.
FAQ
What is digital pathology?
Digital pathology is the acquisition, management, sharing, interpretation, and analysis of pathology information in a digital environment.
In most current workflows, this involves scanning glass slides into digital images and using software to review, manage, share, archive, or analyze those images.
The term is broader than whole slide imaging because it includes the surrounding workflow and software ecosystem.
What is whole slide imaging?
Whole slide imaging is the process of scanning an entire glass slide at high resolution to create a whole slide image.
It is the acquisition layer that made modern digital pathology possible.
The Digital Pathology Association defines whole slide imaging as the acquisition process of creating a whole slide image on a slide scanner.
What is the difference between digital pathology and whole slide imaging?
Whole slide imaging creates the digital slide.
Digital pathology is the broader environment for using that slide, including image management, sharing, review, archiving, workflow integration, and AI analysis.
A scanner may create the image, but digital pathology includes the systems and workflows that make that image useful.
What is H&E staining?
H&E staining uses hematoxylin and eosin to make tissue structures visible under the microscope.
Hematoxylin highlights nuclei in blue-purple tones, while eosin highlights cytoplasm and connective tissue in pink to red tones.
H&E remains the foundational visual framework for much of diagnostic pathology.
Why is H&E important in pathology?
H&E is important because it provides a broad, familiar, high-information view of tissue morphology.
Pathologists use H&E to evaluate tissue architecture, cellular features, inflammation, necrosis, fibrosis, and tumor patterns.
It is often the first slide reviewed and the basis for deciding whether additional testing is needed.
What is IHC staining?
IHC, or immunohistochemistry, uses antibodies to detect specific antigens or proteins in tissue.
It helps pathologists answer questions that morphology alone may not resolve, such as tumor classification, tissue of
origin, biomarker status, and potential response to therapy.
What is brightfield microscopy?
Brightfield microscopy uses transmitted white light to visualize tissue, usually after a stain such as H&E, chromogenic IHC, or a special stain has been applied.
Brightfield is central to routine diagnostic pathology and is also important for emerging Digital Staining workflows that begin with unstained brightfield scans.
What is fluorescence microscopy?
Fluorescence microscopy uses fluorescent labels and specialized imaging systems to visualize specific biological targets.
It can support immunofluorescence, FISH, multiplex biomarker analysis, spatial biology, and other specialized workflows.
Brightfield and fluorescence are different tools for different jobs, not simple substitutes.
What is computational pathology?
Computational pathology applies AI, machine learning, image analysis, and data science to pathology images and related data.
It can support tumor detection, cell segmentation, biomarker quantification, prognosis research, workflow triage, and other applications.
Most computational pathology tools analyze images that already exist.
What is computer-aided diagnosis?
Computer-aided diagnosis, or CAD, refers to systems that assist diagnostic interpretation.
CAD is a narrower category within computational pathology. Computational pathology can include CAD, but it also includes research discovery, quantification, triage, quality control, biomarker prediction, and workflow support.
What are foundation models in pathology?
Foundation models in pathology are large AI models trained on broad pathology datasets and designed to support many downstream tasks.
Recent examples include models trained on very large collections of whole slide images and image tiles.
Their importance is that they may shift pathology AI from individual point solutions toward reusable platform capabilities.
How is PathScience's foundational model different?
Many pathology foundation models are designed primarily to analyze existing pathology images.
PathScience's generalized foundational model is designed around Digital Staining: generating digital stain images directly from unstained tissue scans.
That places PathScience at the image creation and AI platform layer of pathology, not only the image analysis layer.
What is Digital Staining?
Digital Staining is the use of computational methods, including AI, to generate digital stain images from tissue image data.
Some approaches begin with unstained tissue scans. Others may use different imaging inputs.
Because the category is broad, it is important to ask what input, output, tissue type, scanner, workflow, and intended use a system is designed for.
What is Digital H&E?
Digital H&E is an AI-generated H&E image created from tissue image data, such as an unstained tissue scan.
It matters because H&E is the visual framework pathologists already know.
However, visual similarity alone is not the same as validation. Intended use, evidence, morphology, workflow fit, and regulatory context matter.
What is unstained tissue imaging?
Unstained tissue imaging captures tissue before chemical stains are applied.
Unstained tissue may be difficult for humans to interpret directly, but it can provide input data for computational workflows such as Digital Staining or digital H&E generation.
What is Direct-to-Digital Staining?
Direct-to-Digital Staining is PathScience's term for generating digital stain images directly from unstained tissue scans.
PathScience's platform begins with H&E digital image generation and is being developed toward future Digital Staining and intelligent workflow applications.
What is the difference between Digital Staining and Direct-to-Digital Staining?
Digital Staining is the broader category of AI-enabled digital stain generation.
Direct-to-Digital Staining is the more specific PathScience workflow: generating digital stain images directly from unstained tissue scans.
Final thoughts
Modern digital pathology is no longer defined by one technology.
It is an ecosystem that includes tissue preparation, staining, slide scanning, image management, pathologist review, AI analysis, foundation models, Digital Staining, and emerging intelligent workflows.
That is why terminology matters.
A lab director evaluating whole slide imaging has different questions than a researcher evaluating Digital Staining. An investor looking at pathology foundation models is evaluating a different layer of the stack than a pathologist validating primary diagnosis on a WSI system. A technology team building AI image-generation workflows faces different requirements than a team building a slide viewer.
The best way to make sense of the field is to place each concept in the workflow layer where it belongs.
H&E and IHC create visual and molecular information. Whole slide imaging digitizes slides. Digital pathology manages and operationalizes those images. Computational pathology analyzes them. Foundation models may make AI capabilities more reusable. Digital Staining creates digital stain images from tissue image data. Direct-to-Digital Staining, as PathScience uses the term, describes a workflow that begins with unstained tissue scans and generates familiar digital stain images directly.
PathScience's broader ambition is to build the Generalized Foundational Model and AI Platform for Pathology: transforming diagnosis through instant Digital Staining, enhanced precision, and accelerated, intelligent clinical workflows.
Pathology will continue to rely on validated methods, pathologist expertise, and intended-use-specific evidence. At the same time, the field is becoming more digital, more computational, and more connected across workflow layers.
Understanding the language behind that shift is one of the best ways to understand where modern pathology is headed.
Modern digital pathology is no longer defined by one technology.
It is an ecosystem that includes tissue preparation, staining, slide scanning, image management, pathologist review, AI analysis, foundation models, Digital Staining, and emerging intelligent workflows.
That is why terminology matters.
A lab director evaluating whole slide imaging has different questions than a researcher evaluating Digital Staining. An investor looking at pathology foundation models is evaluating a different layer of the stack than a pathologist validating primary diagnosis on a WSI system. A technology team building AI image-generation workflows faces different requirements than a team building a slide viewer.
The best way to make sense of the field is to place each concept in the workflow layer where it belongs.
H&E and IHC create visual and molecular information. Whole slide imaging digitizes slides. Digital pathology manages and operationalizes those images. Computational pathology analyzes them. Foundation models may make AI capabilities more reusable. Digital Staining creates digital stain images from tissue image data. Direct-to-Digital Staining, as PathScience uses the term, describes a workflow that begins with unstained tissue scans and generates familiar digital stain images directly.
PathScience's broader ambition is to build the Generalized Foundational Model and AI Platform for Pathology: transforming diagnosis through instant Digital Staining, enhanced precision, and accelerated, intelligent clinical workflows.
Pathology will continue to rely on validated methods, pathologist expertise, and intended-use-specific evidence. At the same time, the field is becoming more digital, more computational, and more connected across workflow layers.
Understanding the language behind that shift is one of the best ways to understand where modern pathology is headed.
Explore PathScience Direct-to-Digital Staining.
See the technology
© 2026 PathScience. All rights reserved.
© 2026 PathScience. All rights reserved.
© 2026 PathScience. All rights reserved.