Michigan’s pathology department had an edge because 20 years ago, led by Bruce Friedman, MD, emeritus professor at Michigan, the informatics department’s pursuit of software development helped the department self-determine its future capabilities to conduct high-tech ventures involving software development and information technology within pathology itself, says Dr. Balis, who joined the team in 2006.
“This is uncommon among pathology departments, which often witness their IT being subsumed among the enterprise’s central IT group.”
“Michigan’s secret sauce, if you will, is that we’ve retained a substantial critical mass of local IT capabilities, not just for support but also for research and development,” he explains. “So we build our own software, we do original application development, and we do field-testing of candidate applications with our various user communities.”
The creation of a digital primary diagnosis workflow is a multimillion dollar investment for a department of Michigan Medicine’s volume, Dr. Balis says, and justifying it is the potential of AI and machine learning to have a positive impact on the level of care provided to patients.
“Now that the field of pathology is able to gather spatially anchored transcriptomics data—all the molecular data that can now be spatially aligned, if you will, with histology—this will allow for far more precise diagnosis and prediction of the biologic potential for a given individual. So you will be able to look at an image taken in tandem with that exact tissue’s associated pattern of molecular data, put it through one or more computational pipelines, and arrive upon a predictive set of outcomes for a patient’s biological potential for, let’s say, a tumor, to ask what will that tumor do at six months, a year, two years, in terms of its metastatic potential or recurrence potential. And if you compare and contrast that predictive capability with what you can currently do with the use of a microscope alone, there’s no comparison.”
In the U.S., he says, “We’re right on the cusp of the pathology specialty as a whole awakening to the fact that digital workflow for primary diagnosis is transitioning from a nice-to-have to a must-have. It should no longer be viewed as a luxury or a curiosity.” Pathologists should share with all colleagues what works and what doesn’t, he adds, “so people don’t repeat mistakes.”
Mustafa Yousif, MD, a breast and gynecologic pathologist at Michigan and director of digital pathology, points to the importance of the DICOM format for compatibility and integration with other departments, and for flexibility in equipment, the ability to upgrade without compatibility problems, and avoiding the inefficiencies of proprietary formats.
Michigan had a dozen years ago what many pathology departments have now—a standalone IT system, especially for digital pathology, to use for multiple purposes, including research, archiving, education, and consultation, Dr. Yousif says.
The weakness was that the IT has never been available for full clinical diagnosis. “The entire system was only connected to the Department of Pathology. We could not share our images with our colleagues or do multidisciplinary collaboration between radiologists, pathologists, and oncologists to review a case.”