Q&A | The State of Digital Pathology, with Paul Chang, MD
With vendors inking partnerships and developing new tools to bring pathology out of the dark ages and experts pegging the integration of radiology and pathology as one of the top opportunities for imaging, digital pathology may be imaging’s next frontier. This month, Health Imaging & IT chatted with Paul Chang, MD, professor and vice chair of radiology informatics and medical director of pathology informatics at University of Chicago, who offers insights into pathology workflow.

How does digital pathology workflow compare to radiology workflow?

Chang: I’ve been struck by a few assumptions that have been made about pathology workflow. The first is the belief that the pathology imaging task is similar to radiology. People see the success of digital image management in radiology and believe we should apply the same paradigm to pathology. However, pathology workflow differs in fundamental ways which I believe change the direction of digital pathology.

Can you describe these differences?

Chang: As a radiologist, the most important part of my interpretation is whether or not there has been a change from the previous images. I need a high-quality representation of current and prior studies. This is not how pathologists use prior studies.
When evaluating prior studies, pathologists typically refer to a few key selected regions—the fields that represent the prior histology. The pathologist’s job is to tell if the current cells represent the same or different histology. Accordingly, the pathologist almost never is required to review every prior slide.

What are the image management implications of this pathology workflow process?

Chang:The consequences are significant. I don’t have to replicate digital radiology image management, which requires a large, complex and costly archive. It’s hard to justify an archive that is write once, read almost never. Wouldn’t it be better to just capture and store the key regions from the slides? Those selected images would be far easier to store and would be more cost-effective.

That addresses the image storage process, the PACS component. What about the RIS side of imaging informatics?

Chang: This is another fundamental difference between radiology and pathology workflow. In radiology, we acquire the digital image of the patient as one of the first steps in our workflow. From an informatics perspective, we transform the “analog” patient into digital data at the beginning and maintain that digital representation throughout the workflow process, which makes it easier to leverage digital technology to realize quality and efficiency.
In pathology the transformation from physical to digital is not done until near the end of the workflow.  The pathology workflow process requires the handling, tracking and processing of physical specimens throughout. In many labs, specimens are identified with patient demographics by the use of rubber bands binding the specimen container with a paper requisition form. A lot of things can go wrong. The real need here is not digital imaging but specimen tracking.
This very important part of pathology workflow doesn’t need PACS; it needs Fed Ex. Fed Ex manages the handling of physical packages by embracing digital workflow. They use barcoding to make sure packages don’t fall through the cracks. In anatomic pathology, barcoding and similar digital approaches could link the physical specimen to the patient and help manage specimen handling to avoid lost or misidentified specimens.

Can you provide an example?

Chang: At the University of Chicago, we are building a workflow solution that attempts to leverage digital technology appropriately. Barcoding will be used to unambiguously associate specimens with patient demographics and to track these specimens throughout the process. This specimen tracking system will provide the ability to launch contextually relevant decision support tools for users. For example, at the grossing station, scanning the specimen barcode will automatically show the pathologist the patient information and the relevant radiology images to help guide the specimen sampling.
This is good news. We don’t have to wait for PACS archive technology to get significantly cheaper before we can embrace digital pathology. We can leverage existing enterprise PACS to store the much more modest representative images and then apply appropriate digital solutions such as specimen tracking to orchestrate more efficient and safer workflow. We can do this now.