Planning for Digital & AI Pathology? Here Are the 3 Questions Every Clinical Lab Should Ask First

With over 20 years in the pathology diagnostics field, Eric E. Walk, MD, FCAP, is excited to see the rapid momentum and accelerating adoption of digital and AI pathology in recent years.

As CMO of a digital and AI pathology company, Walk has had a front-row seat to the field’s transformation, engaging with pathology laboratories of all sizes to learn firsthand about their digital journeys, strategic priorities, operational goals, and the challenges they face.

Dr. Walk shares his high-level insights on three of the most common questions laboratories have about digital and AI pathology:

What are the key components of a digital and AI pathology system for clinical labs?

The four main components include the 

• laboratory information system (LIS), 
• whole slide scanner, 
• image management system (IMS), and 
• AI applications (see Figure 1). 

The key point here is that a core digital pathology system without AI will enable basic applications and use cases such as primary digital diagnosis, consultations, and education, with AI pathology unlocking additional applications in the areas of workflow efficiency, assisted diagnosis/report generation, and assisted precision medicine biomarker quantification. 

Many of the latter applications are of high interest to laboratories since these represent opportunities to demonstrate return on investment (ROI) through laboratory resource efficiency. 

Should clinical labs implement digital pathology first and add AI later—or implement both simultaneously?

Many labs I speak with are laser focused on digital pathology implementation and want to delay AI implementation to some undetermined future. My advice is to plan for AI from the very beginning, even if the actual implementation comes later. 

The most important factor here is the IMS and making sure that it has the necessary functionality and interoperability to support the AI applications that are valuable for your lab’s workflow. The wrong IMS decision now could lock you out of AI options later and/or create a disjointed workflow, requiring multiple contextual “pop-out” windows vs a more integrated workflow that’s created with IMSs that have natively integrated AI. 

Some of the most compelling AI pathology use cases—such as case prioritization, auto-QC, and smart-case distribution—required tightly integrated AI into the IMS.

How does digital pathology affect pathologist workflow and patient care?

My prediction is that digital and AI pathology will completely transform the histopathology workflow as summarized in the diagram below. A component of this is true “pathology decision support,” where high quality pathology diagnostic skills become democratized globally via a combination of digital pathology, AI algorithm tools, pathology-specific visual language model chatbots and AI agents. The end result will be more accurate diagnoses and in the era of precision medicine, more consistent biomarker testing (e.g., AI-assist tools recommending the guideline approved biomarkers), and better outcomes for patients. An additional benefit could be patient-specific chatbots that translate pathology reports into lay language so that patients are better informed prior to seeing clinicians.

The future of digital and AI pathology

The future for digital and AI pathology couldn’t be more exciting for the pathology and laboratory medicine field. The latest advances in AI technology such as visual language models and AI agents are truly creating novel workflows that can be harnessed to dramatically improve quality and efficiency in ways that have never been possible. It’s vital for all of us in the field to educate ourselves on these new digital and AI technologies so that we can continue to best serve patients and the broader healthcare community.