Understanding the Catalysts Behind the Acceleration of Digital Pathology
Digitization is radically transforming pathology. This change is driven partly by force of circumstance – think pandemic – and partly by the passion of its champions. To find out more, we spoke to David Dimond (Chief Innovation Officer, Global Healthcare - Life Sciences, Dell Technologies) and Michael Valante (Chief Technology Officer, Digital Pathology, Dell Technologies)…
sponsored by Dell Technologies
What triggered your interest in pathology?
MV: I have spent many years working in imaging informatics for healthcare, helping clinical departments and care providers enhance collaboration and patient care. When digital pathology began to emerge, I saw the opportunity to bring my radiology experience to the pathology lab and help pathologists with their digital transformation. Leveraging technology to enhance patient care is highly motivating!
DD: One driver was personal; I had seen family members with cancer having to cope with slow diagnostic and patient management processes. The biggest hurdle was getting all the pathology data in front of different experts at the same time. I wanted to change that. I became convinced that digital pathology was inevitable, and that Dell Technologies – given its impressive technology assets and market share – was the right place to make it happen. The main challenge was integrating different digital healthcare data systems into a unified, vendor-neutral archive, but I believed my 20-year background in electronic health records and medical image digitization would help meet that challenge.
What are the biggest opportunities in the move to digital pathology?
MV: I’m encouraged by the progress being made in pathology toward digital transformation; after all, other medical imaging services, such as radiology, adopted digitization years ago. Now, 15 months of lockdowns have presented a unique opportunity to transform the way pathology uses and shares images and data. COVID-19 resulted in a seismic shift to remote working, which has accelerated the opportunities for digital pathology.
DD: The pandemic has shown everybody what is possible. For example, wide availability of rapid COVID-19 diagnostics has changed expectations; patients now demand faster turnaround. Overall, people are at last beginning to recognize that digital pathology will inevitably become the standard of care and that pathology laboratories must invest in these capabilities, despite methodological limitations and inadequate standardization.
Why should labs make the move to digital?
MV: Removing the geographic limitations associated with traditional microscope-based workflows improves efficiency. More importantly, however, digitization allows pathologists to benefit from more sophisticated technologies – in particular, AI tools. These advanced analytical tools allow pathologists to interrogate data more quickly and reproducibly, thereby optimizing decision-making. I have no doubt that AI will be part of digital pathology’s future.
DD: Digitization releases the power of biobanking and opens up many opportunities, not least of which is in the field of long-haul COVID-19 research – which has attracted NIH funding of ~US$1 billion for projects that include the building of an imaging resource archive for pathology and other specialties. That kind of work demands that you interrogate all the data holistically, so digitization is essential. For example, a new Dell Technologies’ Digital Twin initiative uses technology to leverage digital pathology imaging to identify groups of patients with similar diagnoses and genetic attributes and then applies AI to continuously search for correlations. This approach has great potential, but is not possible with legacy pathology systems.
What barriers exist to digital pathology?
MV: First, it’s essential to have infrastructure capable of managing the large amounts of information and data pathology produces. This requires careful planning, especially regarding interoperability and integration. Second, the technology must be adopted in a way that is compatible with the laboratory’s work processes. In addition to deciding which tools should be used, adoption plans must define initial and growth use cases that will be supported. All these factors can complicate the shift to digital pathology, but they are by no means insurmountable.
DD: The primary barriers are funding shortages and the current status quo, of which the latter can be more challenging. There is still a feeling that “real pathology” is done by humans. But younger pathologists increasingly expect a digital environment and, over time, they will drive change. Another issue might be some IT teams’ reluctance to grapple with sophisticated AI tools and huge digital pathology files. These demand increasingly high-performance computer systems and can’t be managed in the public cloud alone. Finally, we still need to see improvement in standards for genomic data and in the integration of images and associated data in the emerging area of personalized health called “computational pathology.”
How might digital pathology evolve – and how can we future-proof it?
MV: Future challenges include accurate capture and efficient interrogation of data, as well as management of regulatory concerns regarding testing and monitoring. I am convinced that AI will provide solutions to all such issues and will, at the same time, enhance the interpretation of pathology data and its integration with diagnostic and treatment processes. More generally, we should learn from imaging fields where digitization has proceeded further than it yet has in pathology. In particular, we should embrace existing infrastructure solutions and, where possible, adapt them to support pathology’s workflows. Above all, we must maximize the benefits of continuing advances in information sharing to improve clinical collaboration and its resulting improvements in care delivery.
DD: Digitization creates a store of medical information that follows each patient. The key element in this packet of patient-associated information won’t be a glass slide or a tissue block; it will be an image file integrated with other data. This will radically transform our practices. Imagine accessing all available data – the entire virtual patient – via the click of a mouse! These capabilities will facilitate the adoption of AI; we will come to rely on algorithms that are continually fed with data and that not only provide augmented diagnoses, but also support development of breakthrough treatments. For example, vaccine manufacturers will be able to respond to an evolving healthcare situation by modulating their products according to information provided by digital pathology. Importantly, collaboration between different specialists will be enhanced, leading to more holistic clinical decisions. And at the same time, the AI algorithms themselves will evolve. The architectural approach is inherently future-proof!
Any final thoughts?
MV: Seek IT expertise early when planning a digital pathology strategy; otherwise, you may risk overlooking important aspects of the transformation, such as how can we best collaborate now that caregivers are not limited to relying on a physical glass slide and what we can do with the digital data to redesign the current workflows. Digital pathology requires an infrastructure capable of managing huge amounts of data in a scalable, secure, standardized, and reliable way. Such data management is a mission-critical aspect of digital pathology – so the infrastructure that supports it is vital. I make this assertion with confidence, because we at Dell Technologies have digitized data for many different healthcare services; we understand what is required in what can be a challenging endeavor.
DD: Don’t underestimate the expertise required to implement an effective digital transformation in the pathology laboratory. Creating the patient data package requires attention to data validity, reproducibility, and provenance; this is a significant engineering challenge because the data is spread between multiple organizations in multiple clouds. Managing this challenge requires integration of the virtual information packets and establishment of a register of anonymized patient data. Few organizations have the requisite expertise for this; indeed, we see much confusion regarding management and reproducibility of diagnostic information. Fortunately, Dell Technologies has significant experience with this type of challenge. We maintain collaborative relationships with all the cloud providers to ensure that our customers’ technical needs are fully met within the context of clinical workflows, and our skills in this area have proven valuable to many partners.