How Can Image Analysis Help in Digital Pathology?

Stacey Smith, Pharma Tech Outlook | Wednesday, August 26, 2020

Image Analysis in Digital Pathology benefits the healthcare and pharmaceutical institutions in varied ways. Its applications need to be studied, and clinical trials should be performed adequately for better results.

FREMONT, CA: Image analysis (IA) as part of digital pathology has already been around some time. Besides, the first systems were demonstrated in the USA for remote pathologic diagnosis in 1968. A vast amount of advancements in technology have made since then, but the penetration of digital pathology applications in the clinical space remains limited.

There is a range of reasons for this, but one that's very essentials has been compliance with regulatory needs. The pathology IA space for research is dynamic, and various companies are challenging with each other in offering research use only solutions. Many pharmaceutical organizations have their computational pathology divisions, which support their drug development.

At present, the process of drug development is invested mostly in oncology. Some cancer drugs rely on the tissue biomarkers quantified by the immunohistochemistry (IHC) for patient stratification. This is a very qualitative method, for which, semi-quantitative pathology scores relying on visual estimation have been developed.

IHC assay and interpretation is a field for applying IA to make quantification objective reproducible. To altogether avoid adding the variability of pathologist's understanding of the process, comprising of tissue processing,

Even though IHC is a readily available method, standardizing it for IA is a very complex and crucial process, which might not be possible. Both the preanalytical and analytical phases of IHC comprises multiple steps, each of which may contribute to the variability in results and differences in interpretation. This makes it very difficult to keep the assays consistent across laboratories and scale up the availability of the test.

IA algorithms are just the final, quantifying part of the process, usually optimized to a particular appearance. In case if any of these aspects alter, the algorithm will no longer provide correct results. Therefore, the whole process needs to be strictly standardized, which is not always possible when working across laboratories.