UK authorises digital pathology for cancer screening

A new digital pathology technology capable of speeding up the analysis of cancer screening samples has been approved by the UK government based on research carried out by the University Hospitals of Coventry and Warwickshire (UHCW) and the University of Warwick. This will mean faster results, especially for bowel, breast, lung and cervical cancers.

But what is digital pathology? It is the use of automated slide scanners to digitise the histopathological process that examines cells and tissues under a microscope and a fundamental step in the early detection of diseases. Instead of traditional glass slides, samples are scanned to create high-resolution images to be stored, analysed and shared, allowing pathologists and other healthcare professionals to access information and collaborate remotely.

It’s obvious that technological advances have contributed a great deal to the field of pathology, but the concept of digital pathology emerged decades ago, we could say more than a century ago, when images produced by microscopes were transferred to photographic plates for storage. More recently, at the end of the 1960s, scientists began to carry out pathology practices remotely, with access to images produced in laboratories elsewhere. However, in the last 20 years, telepathology has begun to be widely adopted, with scanners for imaging entire slides becoming commercially available – previously, it was only possible to produce images of a few regions of interest. This moment can be considered the turning point for digital pathology, experts say.

As well as making it easier to share samples, digital pathology processes also help to reduce the risk of samples being lost or damaged and reduce the need for pathologists to be present in hospitals, since slides can be analysed remotely. What’s more, researchers predict that the digitalisation of slides will soon enable the use of algorithms to help with analysis, including the use of Artificial Intelligence to identify pathological findings.

Although its widespread application is currently limited by technical issues, infrastructure costs, regulatory standards and aversion to technology among healthcare professionals, digital pathology could be used in the future mainly for teleconsultations. It will also play a crucial role in the modernisation of pathology practices, enabling faster diagnoses, facilitating research and contributing to the worldwide evolution of the digitalisation and connection of healthcare systems.

In the UK, the Royal College of Pathologists, an association with around 11,500 members that deals with issues related to the science and practice of pathology, has a committee working on the development of digital pathology, creating standards, improving practices and raising funds to promote this branch of medicine. The group estimates that the IT infrastructure needed to promote digital pathology in a large UK hospital would cost between £2 million and £4 million, plus, of course, the demand for staff and training for pathologists.

In the United States, the Digital Pathology Association (DPA), founded in 2009, has also been working to ensure standards and interoperability for developers and users of digital pathology technologies.

In Brazil, digital pathology and the use of Artificial Intelligence tools will be among the topics debated at the 34th Brazilian Congress of Pathology and 27th Brazilian Congress of Cytopathology, the largest event in the field in South America, which will take place in Belém, Pará, between 29 May and 1 June 2024.

The digital pathology market is expected to exceed US$3.8 billion by 2034, with a compound annual growth rate of 13.6 per cent by the end of the period.

Use of Artificial Intelligence in digital pathology

The integration of Artificial Intelligence (AI) technologies into digital pathology practices helps speed up previously manual processes and reduce bottlenecks in diagnostic work. The digitisation of images and other data also contributes to information sharing, second opinions and comparisons.

With a view to exploiting the optimisation of Artificial Intelligence models and high-precision inference resources to help pathologists detect and classify biological structures in whole slide images, Wistron, a provider of technological solutions in the areas of communications, components and software, in partnership with Intel, carried out a proof of concept called HepatoAI POC.

The intention is to help organisations that are looking to adopt digital pathology, but need to accommodate this desire within current IT budgets and infrastructures. According to Wistron, the Intel OpenVINO toolkit is a great partner in this endeavour, helping pathologists optimise pre-trained models for inference on general-purpose hardware with Intel processors, iGPUs and dedicated accelerators. The solution can also be used for inference on various devices using industry APIs.

The project used OpenVINO Model Server (OVMS), a scalable, high-performance microservice, to create and deploy pipelines of various models and manage workflows, as well as implement AI models at scale in a production environment, with easy integration into the hospital environments participating in the project.

The solution can also be used for telepathology, i.e. digital pathology at a distance, for example to get a second opinion, carry out training or improve the accuracy of AI models. In this context, Intel Smart Edge can manage several AI models trained in different hospitals within one system. It also only transfers data from AI models, rather than datasets with images of entire slides and CT scans. What’s more, it helps keep models up to date in any number of hospitals within an organisation, eliminating manual processes.