Digital Pathology Workflow Challenges and Solutions

The term pathology comes from the Ancient Greek roots of pathos (πάθος), meaning “experience” or “suffering” and -logia (-λογία), “study of.” The Latin term is of early sixteenth-century origin and became increasingly popularized after the 1530s.

The history of digital pathology stretches back over 100 years, with specialized equipment capturing images from microscopes onto photographic plates to retain, refer to, and share with other scientists, aiding in gaining insights into pathology.

As technology advances and healthcare systems adapt, the path forward for digital pathology is clear – it is not a matter of if but when it will become an integral part of modern healthcare. The journey may be complex, but the destination promises to revolutionize the way we study, understand, and diagnose through the study of pathology.

What is Digital Pathology?

Digital pathology is a dynamic, image-based environment that enables the acquisition, management, and interpretation of pathology information generated from a digitized glass slide. Pathology information can encompass various data types, such as histopathology, cytopathology, clinical pathology, and molecular pathology findings. Whole-slide imaging (WSI) development enables pathologists to “read” digital images on a computer screen instead of a physical slide under a microscope. The first whole-slide imaging (WSI) device, the Philips IntelliSite Pathology Solution (PIPS), was authorized in 2017 for marketing in the U.S. through the de novo pathway. The global digital pathology market in the U.S. was 975.5 million in 2022, with an annual growth rate (CAGR) of 7.7% from 2023 to 2030.

Whole-slide imaging (WSI) scans a complete slide with variable resolution and potentially with different layers instead of just taking a picture through a microscope. The WSI development enables pathologists to “read” digital images on a computer screen instead of a physical slide under a microscope while providing the same functionality.

The concept of telepathology— transmitting microscope images between remote locations — has been around for nearly 50 years. However, in the past decade, pathology has begun to undergo a true digital transformation, moving away from analog into an electronic environment. Limitations in technology have prevented the method from realizing its full potential until the last ten years.

What are the Benefits and Adoption Challenges of Digital Pathology?

The benefits of digital pathology include greater diagnostic accuracy, standardized reporting templates, and improved quality assurance through automated tracking and audit trails. By digitizing tissue samples and leveraging advanced software, digital pathology improves workflow efficiency, reduces operational costs, and enables real-time collaboration and remote consultations among pathologists worldwide. With secure and centralized image storage, easy archiving, and integration with Electronic Health Records (EHR), digital pathology facilitates seamless data management and access. Computer-assisted analysis tools aid in accurate diagnosis and research, while extensive digital slide libraries improve education and training opportunities.

Despite the benefits, the adoption of digital pathology concerns interoperability challenges, cost of implementation, and lagging reimbursement remain a concern in the U.S. Integrating new digital pathology solutions with outdated laboratory information systems and other technologies can be incredibly challenging for smaller labs lacking the resources of larger medical institutions. The adoption of digital pathology in the U.S. has been slower compared to countries in Western Europe and the UK, primarily due to the more fragmented nature of the healthcare system in the U.S. Compared to other countries with unified healthcare systems, such as the UK’s National Health Service, the U.S. healthcare system involves numerous independent hospitals and healthcare providers, making it challenging to implement large-scale technological changes quickly and uniformly.

The economic barrier is justified. However, the digital pathology budget is estimated to be less than 5% of the total budget for a typical pathology department. Compared with other significant IT investments, e.g., deploying an EHR, it is much smaller. The primary driver to start the deployment of digital pathology will be AI. As soon as some of the algorithms are mature enough to improve efficiency and accuracy significantly, there will be an obvious incentive to start the transition. And, of course, being reimbursed for digitizing a slide will also help.

The digitization of pathology is more complex than buying a slide scanner and corresponding image management, processing, and archiving systems. There will be an enormous burden on the infrastructure, especially if the system wants to use the existing network and archiving system, as the data sets are at least ten times the size of the most extensive studies seen in radiology. In addition, many protocols require workflow optimization for the various subspecialties ranging from different fluorescence, resolution, and color display gamma settings at the workstation. Lastly, additional QA steps are needed to ensure the slides are scanned correctly with the correct focus. Some new challenges include ensuring the slides are completely dry before loading them in the scanner so they don’t get stuck, which was not an issue when sliding them under a microscope. And even though most of the pathologists using digital feel comfortable that the system provides at least the same or better diagnosis, there are still a few outliers where a microscope might be preferable. Digital pathology requires a lot of manual labor and specific procedures to generate the pathology exams.

Unlike the adoption of digital pathology, the digitization of radiology images, i.e., transitioning from taking a chest X-ray using analog film to a digital image acquisition, is now mature. Virtually every hospital in the U.S. has a Picture Archiving and Communication System (PACS) in some way or another. We can refer to the radiology digital transformation and its challenges as a model for what needs to happen in digital pathology so we can relate better to its complexity and learn from these early experiences.

Considerations for Implementing Digital Pathology Systems

Bandwidth and Network Infrastructure: Digital Pathology relies on modern technology, IT infrastructure, and cloud computing. Over the last decade, technology has advanced significantly, with 10G network connections becoming widely available and fast SSD drives with affordable large storage capacities becoming more common. This infrastructure is essential for supporting pathology data. The PACS and workflow engines must support a large amount of metadata and specific datasets. The architecture of the pathology workflow in such a way as to avoid affecting the well-established radiology workflow, as well as any other subspecialty workflow that can share the same software and infrastructure with pathology, is crucial. Scanning equipment and software can be expensive when implementing digital pathology systems early. Keeping up with maintenance and maintaining a robust IT infrastructure can strain budgets.

Radiologist Adoption: It takes time to convince radiologists that digital images provide the same or better quality as an analog film so that it would not impact their diagnosis. Image manipulation tools such as changing the window width/level, zoom, measurements, and annotations helped the conversion. Once pathologists try digital pathology, they agree that it offers quality and performance that is at least on par with, if not superior to, traditional pathology images viewed through a microscope. According to one study in 2018, pathologists were 54% and 23% comfortable with rendering a digital primary diagnosis with or without access to glass slides, respectively. But in 2020, these numbers increased to 90% and 60%.


Review of a digital image results on a computer. Source: Philips Healthcare

Image presentation: Over the years, medical grade monitors used for radiology have become stable because of their real-time auto-calibration, show images consistently by using the greyscale representation and standard mapping of pixels into luminance as defined by the DICOM (Digital Imaging and Communications in Medicine) standard and provide the required resolution which is getting closer to the actual pixel resolution of the digital image. We can depend on this technology for pathology as high-resolution color monitors, which show a larger field of view than can be seen through a microscope, are now available.

Image quality: Color calibration of the monitors using ICC profiles for color consistency occurs at the workstations. FDA-approved monitors, especially for digital pathology, are available. Reports have shown that color Gamma corrections at the monitor were needed for specific cases to provide diagnostic-quality images, but overall, these issues have proven to be manageable.


A Pathology image displayed on a computer Image Source: Barco

Telepathology: As with radiology, the COVID-19 pandemic significantly impacted remote interpretation. Instead of shipping slides around using your favorite carrier, the slide image source can keep the original physical slide and share the digital copies with the physician for interpretation. As with any other imaging specialty, remote pathology consultations have drastically increased.

Teaching, education, research, and collaboration: Lecturing to a group of students while they all have access to the same digital image, face-to-face or remotely, is invaluable. Creating a digital teaching file is much more practical than having a set of physical slides around. Having a repository available on digitized sides is critical for research purposes, clinical trials, and to provide training data sets for AI algorithms. For example, NIH provides a Covid digital pathology repository, and several academic institutions (Mayo and others) are making their library of digitized pathology images available.

“There are a lot of similarities between the digital transformation that happened in radiology and will need to happen in pathology. Key differences are ROI, workflow, and orchestration.”

Patient-centered access: A radiology diagnostic workstation is synchronized with an EHR for that patient to provide a patient-centric view. Many institutions log into their EHR to see the complete patient folder, which might include notes, lab results, history, and any other information available, and then access the images. Similarly, as a pathologist looks at a digital slide, they can simultaneously review any other pertinent information about the patient, which is typically shown on an additional monitor and synchronized with the study, meaning that when there is a context change between the patients in the pathology display, it automatically changes the respective patient information.

Workflow and User Adoption: The pathology workflow differs significantly from radiology. Implementing digital radiology eliminates the steps associated with film processing. However, with pathology, we still need the physical media, i.e., the slide, and we need to add at least one step for the scanning and potentially other steps for the QA of the digitized image. That is why downstream efficiency improvement using AI is critical to making a valid use case. Transitioning from traditional glass slide-based to digital pathology workflows requires change management and user adoption strategies. Training pathologists and healthcare professionals on new technologies, ensuring workflow efficiency, and overcoming resistance to change can be challenging but necessary for successful implementation.

Integration with the Laboratory Information System (LIS): The Accession Number is a crucial link between the scanner and LIS, typically using a barcode printed on the slide. The same link applies in radiology, where the EHR or RIS will assign a number to the order unique within the department and used to link the order with the images and report. When scanned, a slide will use this Accession Number to associate the order and patient demographic information so that the digital image can be tagged appropriately with that information, as the PACS will need that to update its image manager/database.

Integration with the reporting system: Some radiology modalities make extensive measurements, which are exchanged using DICOM Structured Reports and interpreted by a reporting system, and specific fields are auto-populated in the diagnostic report. These measurements, such as for ultrasound, are usually done manually but will increasingly be automated using AI tools. Similarly, Pathology encodes results in machine-readable form, exchanged with the pathology reporting system, and ingested. As these progress, this innovation is where digitization will have a significant impact.

Integration with an Electronic Health Record (EHR): Physicians are increasingly moving from departmental access and perspective to patient-centric access. Integrating pathology images and associated data with the EHR system ensures access to the comprehensive patient record, enabling easy access, data sharing between physicians and specialists, and decision support across the healthcare enterprise.

Storage Infrastructure: One could use the Vendor vendor-neutral archive (VNA) on-prem or in the cloud, but the additional bandwidth and storage requirements are significant. For reference, Mammography studies, currently the most extensive radiology studies in size, can range from 450 MB to 3 GB, while Digital Pathology images start at 1GB. An annual production of a mid-size hospital that creates 50-100TB of radiology imaging data could produce .5-1PB of digital pathology imaging. There is a trend to archive all medical imaging in the same archive and make it available on a single enterprise imaging viewer for physicians; however, one needs to ensure that the current PACS system is sufficiently scalable.

“Standards support, especially the DICOM WSI format, is abysmal and will hamper provider integration and best-of-breed selection.”

Standards: In radiology, the DICOM (Digital Imaging and Communications in Medicine) standard significantly pushed for interoperability between radiology devices and other specialties such as cardiology, dermatology, surgery, etc. Unfortunately, even though there is a DICOM standard for Whole Slide Imaging (WSI), the adoption among most scanner manufacturers has been slow. Vendors must implement a common data standard for these devices to communicate successfully to achieve true digital pathology interoperability. Without a vendor-neutral pathology imaging format, these devices are siloed in labs worldwide, creating massive workflow bottlenecks. As a result, DICOM established Working Group 26 (WG-26) to expand the DICOM standard to pathology imaging using open, accessible, and publicly available standards. One significant differentiation between a photographic image, as supported by DICOM Visible Light (VL) photography for many years, and WSI is that it scans the entire slide, layer for layer, allowing a pathologist to simulate as if using a microscope. In addition to company proprietary formats, other image formats such as tiff or jpeg are common. One pathologist recently commented at a conference that he had to deal with 13 different formats. In addition, simple “screen saved” images provide a snapshot as a DICOM Visible Light microscopic image or a Secondary Capture instead of delivering the WSI DICOM format. Unless an institution wants to stick with a single vendor and use proprietary formats, it is not always the best choice, as there is a barrier to replacement and inter-hospital exchange. There is a need to translate proprietary scanner formats from existing scanners and potentially archived images in a proprietary format. From a standards support perspective, pathology is trailing far behind other imaging specialties, especially in supporting DICOM.

Workflow challenges and how to address these with Digital Pathology:

Instead of moving around physical slides and ensuring they end up at the desk of a pathologist, the digital pathology image management infrastructure will have to take over this task. However, there is much more involved than just routing the output of the slide scanners to the physicians, especially when considering the QA steps and possible plethora of AI algorithms that need to be applied. These are the additional workflow orchestration steps that are required:

Managing the digitization of existing slides: The digitization of the pathology department will typically kick off with massive digitization of existing slides. Some institutions in the US are already starting this process in anticipation of converting to digital. Especially for an academic institution, the first step would be to use digitized images for lectures and educational purposes and conferences such as tumor board meetings. We should be aware of this effort, as digitizing the existing slide library can be time-consuming, potentially spanning many months or even years. Linking the slides to the right patient and study is vital; connecting with the Laboratory Information System is critical to match the image with the right patient. As in radiology, each pathology procedure has an Accession Number, which should be unique within the department and can be used to retrieve the order and obtain patient demographics so the images can be indexed correctly in the image management (PACS) database. The Pathology Image management can use the existing PACS and VNA infrastructure, but be aware that there will be a massive increase in image storage and performance requirements. Numerous institutions are adopting cloud archiving and uploading images for seamless scalability, contingent on the cloud connection’s bandwidth meeting the necessary throughput. Developing a comprehensive plan to address the challenges associated with digitizing slides is crucial. This plan should involve assessing the current infrastructure, identifying necessary changes, and strategically outlining the digitization process, recognizing that it may demand significant time and effort.

“In addition to the massive digitization effort of the existing slide inventory, orchestration of the workflow to and from AI applications and specific pathology-related QA steps introduced with the scanning attempt is critical.”

Managing the conversion of various non-DICOM digital formats: Some institutions might have a digital teaching file or scanner for tumor board conferences and telepathology. As mentioned earlier, DICOM WSI image formats are rarely supported. Hence, it is essential to incorporate the conversion of images into the DICOM Whole Slide Imaging (WSI) standard as a component of the institution’s comprehensive image management and archiving solution. To tackle this issue, the institution should take the following steps: conduct an inventory of the current equipment and interfaces and formulate a plan that includes upgrading or replacing them, potentially integrating additional middleware for format conversion.

Orchestrating the AI inputs: Depending on the case, images will have to be routed to different algorithms to detect specific findings, e.g., a hematology slide will be subject to a different algorithm than a biopsy of a tissue looking for possible breast cancer. If the AI algorithm is deployed at the workstation or on the PACS, it may not necessitate additional workflow considerations. Nevertheless, to address this orchestration, one can anticipate the deployment of multiple AI algorithms on an AI platform, either on-premises or in the cloud. Additionally, anonymization may become necessary if the images are processed off-site or in the cloud.


Source: ITN Online

Orchestrating the AI detection results: The AI detection results must be rerouted back to the PACS to aid pathologists in interpreting. Furthermore, it’s reasonable to anticipate that the interpreted results will be sent to the reporting system and included in a report without human intervention. Please note that AI for pathology is still very much in its infancy; of the more than 500 approved AI algorithms for imaging, only 4 deal with pathology and 4 with microbiology. We still have a long way to go and a lot of catching up to do. One might assume the “chicken and egg problem:” We need available data sets to train an algorithm on digital datasets. Even though those datasets are available in several European countries that started the digitization process 5-10 years ago, they might represent a cultural and ethnic bias compared with the US. Orchestrating the results also requires the deployment of a platform to be effective.

Orchestrating the QA results: There is an opportunity to have AI detect image quality issues, with out-of-focus being the most common cause. In specific scenarios, the scanner may trigger an automatic rescan or automatically generate an order for additional staining and grossing (i.e., make another cut from the specimen). Metadata scanning and tag morphing capabilities are essential for triggering rescans and comparing metadata information with actual datasets. An intelligent router or AI platform will handle this orchestration.

Workflow Standardization and efficiency. The variations in practices and protocols make it challenging to create standardized workflows for digital pathology for application across different institutions and settings. While digital pathology has the potential to streamline workflows, it can also introduce bottlenecks if not implemented properly. Tailoring the viewer and modality worklist systems to the lab’s specific needs is essential for maximum efficiency.

Orchestrating the worklists, aka intelligent case distribution: Depending on the case, based on AI findings, priority, and for load balancing or sending to specialty pathologists, images are added to the respective worklist of the specific pathologist.

What’s Next In Digital Pathology

The journey of digital pathology is rooted in the ancient origins of the term “pathology,” evolving over the centuries into a field that has now embarked on a transformative digital revolution. While the history of digital pathology spans more than a century, it is in the past decade that we have witnessed significant advancements in technology, network infrastructure, and user adoption, setting the stage for its widespread adoption.

The path to realizing the full potential of digital pathology has its challenges and roadblocks. Bandwidth and network infrastructure, user adoption, image presentation, quality, and integration with various systems are crucial in successfully implementing digital pathology. Telepathology, teaching, research, and collaboration have seen a surge in importance, especially considering recent global events like the COVID-19 pandemic.

One of the driving forces behind the adoption of digital pathology is the promise of artificial intelligence (AI) and its potential to enhance efficiency and accuracy in diagnosis. As AI algorithms mature, the incentive to transition to digital pathology becomes even more apparent.

However, it’s essential to acknowledge that the digitization of pathology is more complex than acquiring scanners and software. It involves a complex orchestration of workflows, including managing the digitization of existing slides, converting various non-DICOM formats, and integrating AI and QA processes. Standardization and workflow efficiency remain challenging due to variations in practices and protocols across institutions.

The digitization of pathology holds great promise in improving diagnostic capabilities and advancing research and education in the field.

Dicom Systems and Digital Pathology Workflows

Dicom Systems partnered with one of Switzerland’s premier healthcare institutions to deploy enterprise imaging workflows for faster transmission of extensive Pathology studies for interpretation. This collaboration enabled enterprise-wide imaging workflows, streamlined data exchange, and enhanced patient care by expanding enterprise imaging operations into digital pathology.

By integrating Dicom Systems’ advanced enterprise imaging solutions, this organization aims to achieve seamless interoperability across its diverse imaging systems, including routing large pathology images and neutralizing bottlenecks and latency challenges. Pathology images are massive objects, on average ten times the size of a mammography image, requiring reliable solutions to route them efficiently and effectively.

The Dicom Systems’ Unifier® platform offers robust capabilities for handling large image files, ensuring efficient routing and storage of pathology images across infrastructures, enhancing diagnostic accuracy, and improving patient care. Contact a workflow expert to learn more about how Unifier can help you achieve greater efficiency and productivity for digital pathology workflows or to discuss a specific use case.