Pathology Experts Reinforce the Importance of Incorporating Measurement Science in Digital Pathology Workflows

by Christos Evangelou, MSc, PhD – Medical Writer and Editor

The benefits of digital pathology are increasingly being recognized by healthcare providers and scientists, especially since the beginning of the COVID-19 pandemic. Digital pathology workflows can increase throughput, accelerate diagnosis, and facilitate remote examination of tissue slides. Nonetheless, the lack of standardized workflows and the use of proprietary data formats and black-box analysis software remain key barriers to the widespread adoption of digital pathology in diagnostic laboratories.

The National Measurement Institutes (NMIs) comprise experts in developing standards and best practices in various domains, including healthcare. To understand the role of standards and measurement science in digital pathology, the National Physical Laboratory (NPL), the UK’s NMI for measurement science, has launched an interdisciplinary project, “Metrology for digital pathology to improve patient outcomes.”

In a recent study that opened the project, NPL, together with a group of pathology experts, analyzed how variations in sample handling, data interoperability, equipment calibration, and image processing act as barriers to the clinical adoption of digital pathology. The results of the study suggest that incorporating measurement science into existing digital pathology workflows can improve current practices in digital pathology and increase confidence in digital pathology-based diagnosis.1

“Despite the growing interest in artificial intelligence tools, molecular imaging, and other novel imaging techniques, the lack of standardized guidelines slows their use for clinical purposes,” said Marina Romanchikova, leader of the digital pathology project at the NPL and first author of the study. “The use of vendor-neutral interoperable data formats, standardized calibration tools, and reproducible laboratory workflows and image analysis software is needed to ensure that digital pathology is trustworthy.”

Dr. Romanchikova also noted that measurement science, including pathology-specific metrics, uncertainty estimation, and equipment calibration, is key to ensuring the reproducibility and high performance of digital image analysis using artificial intelligence (AI) and increasing confidence in diagnosis. “All of the above need to be accompanied by an upskilling of the pathology workforce,” they added. The study was published in the Journal of Pathology Informatics.


Approach: Reviewing and Discussing Existing Evidence

To better understand the challenges arising at different digital pathology workflow stages, NPL has reviewed existing evidence and conducted an online survey of pathology experts.1 A workshop with representatives from healthcare, academia, the pharmaceutical industry, regulatory bodies, and software manufacturers was used to build consensus on the needs and interventions required to improve tissue processing, equipment calibration, image analysis, and data interoperability. The full findings of the study are collated in an NPL open report.2

“This study is the first attempt to evaluate the role of NMIs in the UK digital pathology landscape. We aimed to enhance the existing body of evidence by collating the views from a heterogeneous community of pathology practitioners, equipment vendors, regulators, and researchers,” Dr. Romanchikova said.


Identifying Gaps in Digital Pathology Workflows

The majority (80%) of pathology experts identified the lack of data interoperability and issues with data integration as critical barriers to the efficient navigation of digital pathology data.1 All workshop attendees agreed that integrating imaging and non-imaging data is a key challenge in using digital pathology for diagnosis, and 90% stressed that lack of training and performance metrics hinders pathologists from harnessing the benefits of AI and machine learning for digital pathology.

“While several whole slide imaging (WSI) vendors offer off-the-shelf integrated pathology systems, the interoperability is lost as soon as the data leaves the system,” Dr. Romanchikova explained. “All participants stressed the importance of integrating imaging and non-imaging data for diagnosis. A vendor-neutral data standard such as DICOM is a ‘must-have’ for enabling data collection and analysis.”


Metrology Can Improve Digital Pathology Workflows

Improved equipment and software calibration and traceability were identified as essential for establishing harmonized and reproducible protocols for tissue processing and image acquisition. “We noted the lack of consensus on the scope, methodology, and frequency of calibration procedures for WSI, leading to significant variations in results and affecting the downstream analysis,” Dr. Romanchikova noted. “Tools and methods for calibration and providing traceability were seen as essential to establish harmonized, reproducible sample processing and image acquisition pipelines.”

Adopting metrics for uncertainty, fitness-for-purpose, and reproducibility is needed to ensure the interpretability and trustworthiness of WSI analysis, especially when AI and machine learning tools are used. “Although good laboratory practice and good clinical practice are used in many laboratories, they need to be enhanced by fit-for-purpose quality metrics and standardization routines for tissue staining,” Dr. Romanchikova explained.

Although vendor-neutral interoperable data formats for WSI can provide a common language for exchanging data with other clinical systems, DICOM alone does not guarantee data quality. The pathology community needs “robust metadata standards to capture sample handling, imaging device setting, and image pre-processing steps,” which should be captured using consistent clinical terminologies, ontologies, and units of measurement.


Metrology Can Improve Adoption of AI and Novel Imaging Techniques

All participants were interested in the use of AI and machine learning to support various stages of the digital pathology workflow, including image quality assurance, image analysis, annotation, review prioritization, and disease classification. However, the adoption of AI tools is limited by large image sizes, image artifacts, color variations, and regulatory approval.

“Access to large well-annotated datasets and protocols for training and validation of algorithms are needed to improve the interpretability and explainability of results,” Dr. Romanchikova noted. “We believe that the NMIs, with their wealth of multi-disciplinary expertise in various areas of measurement, are well-placed to develop reference methods and standards for digital pathology and biosciences.”

In addition to WSI analysis using AI, there is growing interest in molecular imaging modalities, such as super-resolution and light sheet microscopy, Raman spectroscopy, and mass spectrometry imaging. “Training and guidelines on the use of these modalities, as well as software tools to integrate these data with WSI, was seen as a medium-to-long-term priority,” Dr. Romanchikova said. To aid pathology practitioners in navigating the complex landscape of molecular imaging, NPL has recently published a comparative review of the technical capabilities of contemporary molecular imaging techniques and the associated needs for data processing.3


Looking Ahead

Although this study was the first to evaluate how the metrology approach can help guide data capture, data analysis, and WSI calibration, more comprehensive guidelines are needed to standardize other aspects of the digital pathology pipeline and to support the uptake of digital pathology in the clinic.

“We intend to focus the initial efforts on working together with digital pathology practitioners and national/international standardization bodies such as Integrating the Healthcare Enterprise, DICOM community, and British Standards Institute to advance the development of platform-agnostic digital pathology data standard,” Dr. Romanchikova noted.



  1. Romanchikova M, Thomas SA, Dexter A, et al. The need for measurement science in digital pathology. J Pathol Inform. 2022;13:100157. doi:10.1016/j.jpi.2022.100157
  2. M. Adeogun, J. Bunch, A. Dexter, et al. Metrology for Digital Pathology. Digital pathology cross-theme project report 2021. [Internet] Accessed Dec. 2022. Available from
  3. Dexter A., Tsikristis D., Belsey N. et al. Next Generation Digital Pathology: Emerging Trends and Measurement Challenges for Molecular Pathology. J. Mol. Pathol. 2022, 3(3), 168-181;

Share This Post

Leave a Reply