The need for measurement science in digital pathology

Area of Science:

• Digital pathology
• Medical informatics
• Measurement science

Background:

• Pathology services faced increased demand during the COVID-19 pandemic.
• Digital pathology offers increased throughput but faces challenges with non-standardized workflows and proprietary software.
• Existing digital pathology solutions often yield data of variable quality due to black-box processing.

Purpose of the Study:

• To present UK expert views on barriers to digital pathology adoption.
• To identify the necessary input of measurement science for improving digital pathology practices.
• To support the UK's digitalization efforts in pathology services.

Main Methods:

• A review of existing evidence on digital pathology.
• An online survey distributed to domain experts in pathology.
• A workshop involving 42 representatives from diverse sectors including healthcare, industry, and academia.

Main Results:

• Lack of data interoperability is a major barrier (80% of attendees).
• Integrating imaging and non-imaging data is crucial for diagnosis, with 80% prioritizing data integration.
• High interest in AI/machine learning (90%) but a need for training and performance metrics was identified.

Conclusions:

• Digital pathology requires interoperable data, reproducible workflows, and trustworthy analysis software.
• Adoption of novel techniques like AI is slowed by a lack of guidance and evaluation tools.
• Measurement science can enhance reproducibility, comparability, and data quality, increasing diagnostic confidence.