By Rob Monroe, MD, PhD, Chief Medical Officer, Leica Biosystems, & Chief Scientific Officer – Oncology, Danaher Diagnostics

In this article, Rob Monroe details Sweden-based Inify Laboratories experience with development of a fully digital, AI-supported workflow to enhance timeliness, consistency and quality of cancer diagnoses. Diagnostician and educator Claes Lindh believes precise, clear and reliable image quality is paramount to maximize diagnostic accuracy for pathology and to optimize patient care.

“Something I have often thought about in my career is how to provide diagnoses that are both precise and efficient,” said Dr. Lindh, a pathologist who has worked with leading institutions such as Karolinska University Hospital and currently serves as Chief Medical Officer of Inify Labs. Inify is pioneering a new approach to pathology services which applies digital tools and artificial intelligence to routine histopathology to improve efficiency and precision in pathology.

Based in Sweden, Inify relies on a fully digital, standardized and AI-supported workflow to optimize quality and response times. “Prostate cancer is our initial focus, and the approach is structured to work at scale to address other cancer types,” says Lindh, who specializes in urological and gynecological cancers. “Our service encompasses the whole chain of sampling handling: from logistics to sample preparation to reporting by a pathologist. In large part, we were compelled to do so to ensure we collect the quality of images we need to get the outcomes we want.”

Preparation of a histological slide is a complicated, multi-step process. “Each step can be done a bit differently and with different chemicals, resulting in variation in stain quality and appearance,” explains Kristian Eurén, Inify’s product manager. “Depending on the laboratory process, the image for the glass slide of each histological specimen will vary somewhat over time in the laboratory and between laboratories. These differences in staining can result in differences in pathologist interpretation – for example, tumor grading and staging – which impact patient care.”

While variation between pathologists is well recognized and documented, what is less known is the impact of variation in image quality. “When pathology is practiced with standard microscopes, it’s difficult to compare and evaluate staining differences between labs and over time at the same lab. Digital pathology offers new opportunities to help pathology labs achieve more consistent stain quality and less variation – and ensure higher quality pathology diagnosis for all patients – by standardizing laboratory workflows and incorporating image QC measures to optimize image quality,” continued Eurén.

Widespread availability of increasingly better digital pathology tools should help ensure reliably digital-ready slides and standardize workflows, Dr. Lindh asserts. “Digital technologies can simplify many of the traditional steps in histopathology so pathologists can focus on critical diagnostic decisions. This is a big change that should involve review of the entire anatomic pathology workflow to standardize it – starting with the handling of biopsies and surgical specimens through the detailed laboratory steps of tissue and slide preparation, diagnostic evaluation, and reporting.”

Dr. Lindh’s work allows him to apply these insights to his practice. Among other innovations, Inify has developed a clinically-validated algorithm to aid pathologists with prostate cancer diagnosis. In developing the algorithm, the Inify team gathered thousands of images from more than 40 hospitals. Analysis of the extensive, robust data set highlighted the challenge of variable image quality on pathologist annotations and diagnosis.

“It reinforced to us how very important it is that images have consistent quality and look the same every day, every time,” Dr. Lindh explains.

To achieve this consistency, Inify sought to optimize the workflow specific to image preparation, recognizing that “digital pathology is much more than buying equipment. It’s also a matter of evaluating processes, educating staff, and standardizing the entire histology process,” Eurén noted.

For example, Dr. Lindh noted  that the lab’s digital slide scanners, Leica Biosystems’ Aperio GT 450 DX, underwent extensive validation. In their validation report, Inify documented that the scanners were easy to use, performed reliably with high sensitivity and specificity, and demonstrated low instrument-to-instrument variability. Further, the scanner validation revealed additional opportunities for standardization in the sectioning and staining steps of slide preparation prior to  digitization.

Dr. Lindh and the Inify team continue to refine their workflow and are optimistic about digital pathology’s potential to improve pathology diagnosis and patient care worldwide. “Digital pathology and AI enable pathologists to become more efficient and accurate in their case reviews and thereby to increase their diagnostic confidence.”