Artificial intelligence (AI) applications hold the promise to transform the detection and diagnosis of diseases. Offering the potential for increased efficiency, accuracy, and faster screening for conditions such as breast, bowel and cervical cancer. As pathology departments across the country explore the digitisation of slides, integrating AI into their workflows offers the chance to rapidly enrich and improve diagnostics.
However, the transition to AI-powered pathology comes with significant security and governance considerations that must be addressed to ensure safe and effective implementation.
AI is already being utilised in various connotations across healthcare, from large language models and generative AI to research applications. In pathology, AI can be particularly transformative, offering the capability to analyse large volumes of data and identify patterns in histopathology samples that indicate the presence of cancers or other diseases. Traditionally, moving glass slides is costly, delays diagnoses, compromises security and limits collaboration. The potential for AI to assist in clinical settings, working alongside human pathologists, is vast, ranging from routine analysis to more complex diagnostic tasks. The technology can support early diagnosis of critical illness, with results delivered in a matter of hours rather than days or weeks.
The case for AI integration
The current state of pathology in the UK highlights the urgency for AI adoption. The NHS aims to achieve a 75% diagnosis rate for cancers within 28 days, but as of April this year, over 70,000 patients waited longer than this for their diagnosis to be confirmed. Additionally, a large proportion of UK pathologists are aged over 55, and 60% are set to retire within the next five to ten years. When you couple this with the increasing demand for cancer services, the case for AI integration becomes stronger and highlights a need to enhance diagnostic efficiency and capacity – while reducing complexity for pathologists.
One of the most promising applications of AI in pathology is the analysis of histopathology samples to detect patterns based on specific biomarkers. This can lead to earlier and more accurate diagnoses, significantly reducing the time it takes from biopsy to results. AI-based image analysis algorithms can automate many time-consuming tasks, enabling laboratories to work more efficiently. However, despite these benefits, widespread adoption of AI in pathology has been slow, mainly due to the costs of implementation, calibration, and ongoing maintenance.
The digitisation of pathology is still in its early stages in the UK, presenting a significant opportunity to accelerate diagnosis times and improve patient outcomes. With a faster diagnostic pathway, patients have more treatment options, the financial impact is reduced and ultimately, these small gains save lives. As pathology departments begin this transformation, it is crucial to consider how AI can be integrated in a secure and effective manner.
Bringing AI to the data centre
A key consideration is where to process these data. It is generally more secure and efficient to bring AI capabilities into a centralised data centre rather than distributing data across multiple external systems and suppliers. Centralised data processing reduces security risks and ensures greater control over data governance and business processes. Whereas a distributed model requires the careful selection of AI suppliers and a robust ecosystem management strategy to handle multiple agreements and data processing protocols.
To adopt AI safely and securely, pathology departments must address several critical factors. First, limiting exposure by minimising the number of external AI systems involved can protect against vulnerabilities.
Centralising AI within a data centre provides a controlled environment where data governance and security protocols can be more effectively enforced. This approach also mitigates the risks associated with having data flow into different companies, each with its own architecture and security practices. In the event of an application failure, a centralised system offers greater resilience and reliability.
The importance of adopting AI safely and securely cannot be emphasised enough. Recent cyberattacks targeting NHS pathology testing organisations have clearly demonstrated the disruption that can be caused through security breaches, and the impact this has on operations and outpatient appointments. By maintaining control over AI applications and data within a centralised system, pathology departments can better protect against such threats.
A second pair of eyes
AI is not intended to replace human pathologists but to expand their capabilities. It has the potential to serve as a ‘second pair of eyes,’ automating and recognising specific biomarkers and analysing large sets of slides. AI’s role is to support the rapid identification and prioritisation of cases, provide second opinions and enable pathologists to focus on more nuanced diagnostic decisions.
As pathology departments begin their digitisation journeys, the integration of AI offers a promising path forward. However, careful planning and consideration of security and governance issues is essential – as is collaboration between technologists and pathologists. Not only to ensure that AI algorithms can learn and adapt from real-world feedback but also to make sure they are reliable and integrated into clinical workflows. This will not only enhance diagnostic accuracy and efficiency, but also lead to improved patient outcomes, helping to set a new standard for the future of pathology.
Jonathan Bridges
Chief Innovation Officer, Exponential-e
About Jonathan Bridges
Jonathan brings experience from all spectrums of the IT lifecycle from engineering, design and development, strategy, consulting, service delivery and transformation. At Exponential-e his core focus is supporting customers in transforming their IT services to enable the business to drive their digital transformation strategy. www.exponential-e.com