Most of us aren’t aware of the full extent in which digitization and AI technologies can impact the histopathology lab. However, over the last 25 years, we’ve seen a period of extraordinary development of digital pathology in diagnostic research and drug development.
In our recent webinar, Senior Pathologist Dr. Thomas Lemarchand and Digital Pathology Manager Derick Vollmer take an in-depth look at the growing use and benefits of digital pathology in the field of pharmaceutical research and development.
We’ve compiled the webinar’s key takeaways into this 2-part blog series. You can watch the webinar, “Evolution of AI and Digitization in Precision Pathology and Image Analysis,” on demand here. Or, keep scrolling for an introduction into digital pathology in part 1 of our 2-part blog series.
What are the digital pathology specifics in R&D pharma?
As Dr. Lemarchand explains, be it diagnostic (“clinical”) pathology or experimental (“research”) pathology, AI can increase the speed and accuracy of analyses, accelerate reporting, and ultimately remove time-intensive tasks from a pathologist’s workload, such as scoring.
Regarding pharmaceutical R&D, digital pathology usually involves the scanning and digitizing of thousands of glass slides for a single large chronic study, or spanning several smaller studies. AI tools can help pathologists rapidly analyze these slides (once digitized) and navigate the multiple variables and the need of quantified data. For example, AI tools can assist research pathologists with the scoring of a lesions, especially with the lesions which may exist as low-level background in control untreated animals, to determine a statistically significant threshold for reporting a significant adverse lesion and what is the dose at which it appears. It is especially important in target organs when a safe dose of a treatment modality needs to be found (as is the case in a safety study).
Because experimental research pathology often involves control samples, AI and machine learning methods can also be utilized in the finding of abnormal tissues since the control samples provide a significant, not to say huge normal baseline to compare against. Additionally, once an algorithm has been developed and validated in previous studies, experimental research pathologists can bypass scoring altogether and advance directly to quantification. Indeed, a developed and validated algorithm will be far more precise and eliminate inter-observer imprecisions or subjective bias.
How digital pathology can help treat disease
Through digital pathology, scientists may be able to better treat diseases such as cancer, neurodegeneration, and rare genetic diseases. Through integrations across several disciplines, digitization and AI technologies provide the foundation for precision medicine, which enables the stratification of patients to know whether they will respond to specific treatments. New therapeutic modalities are expensive but effective—and AI can help to make sure only the patients who would benefit from such modalities are treated. Digital pathology is striving and should continue to strive within this context.
One AI-enabled technique for such patient stratification is known as computational pathology, which involves combining and analyzing multiple patient datasets (clinical information, live imaging data, pathology information, histopathology image data, and meta-data) to extract patterns and understand features. While not yet fully automated, computational pathology will certainly streamline the pathology workflow, accelerate drug development, and ensure patients receive the treatment with the greatest odds of success.
Learn how precise “precise” pathology is
Watch the on-demand webinar to learn about precision pathology in greater detail. You can also read part 2 of our blog series for a more in-depth look at digital pathology workflows, the benefits of digital pathology, and how to manage digitized images and other information efficiently.
Do you want to learn more about the advancements in digital pathology AI tools? Contact StageBio to learn more.