PACS embraces AI

Given the many new developments related to the ePA or the German Hospital Future Act, you could be forgiven for thinking that advances in radiology, including PACS, were taking a backseat. The truth is actually just the opposite: radiologists are beginning to view the process of interpreting images in a whole new way, with the potential to greatly simplify traditional workflows. And VISUS has a powerful role to play in that development.

Likely the most exciting issue when it comes to interpreting radiological data is the possibility of incorporating artificial intelligence (AI), or better still: software solutions based on artificial intelligence that simplify the complex process of analyzing images. The strategy that VISUS is pursuing here is based on—what else?—the use of standard formats that the AI software employs for transferring information. As always, the major advantage of using standards consistently is that all AI software solutions can potentially interact with the PACS.

Data transport rather than software integration

VISUS is focused on receiving, visualizing and having the ability to process external data. This also eliminates the need for integrating AI solutions that would serve only to limit the flexibility of JiveX. Plus, JiveX remains lean, user-friendly and straightforward without having to forego the added value that artificial intelligence delivers. JiveX users experience AI results as an interactive overlay that they can scroll through and navigate as usual. The primary difference here is that physicians receive additional information on the location of lesions and on the likelihood that these are malignant. This layer is not mandatory, however, can be toggled on and off as desired. VISUS is currently working on prototypes for interpreting mammography and thoracic imaging—areas where AI applications are already well advanced. It will take a few more months before the first JiveX users can use AI to optimize how they interpret image data—the issue of AI is currently progressing at a rapid pace, however, suggesting that we can expect some developments during this period. Actually integrating AI applications into the interpreting process is a bit farther off in the future, but work in this field is likewise in high gear.

PACS-controlled interpretation workflow and structured interpretation

Speaking of the interpreting process, there have been some practical developments here over the past few months as well: integrating voice-recognition software now allows radiologists to interpret images within JiveX, so that they no longer have to switch back and forth between the viewer for studying an image and the RIS for reporting. All other processes typical of an RIS, such as managing patient files and appointment scheduling, remain in the RIS. This meaningfully separates the physician's job of interpreting images from the radiographer's job of conducting the examination. What is known as the “PACS-driven workflow” has already become accepted practice in the Netherlands—often a pioneer in medical processes. While systems like these also provide a structured platform for interpreting images, very few radiologists today take advantage of report forms tailored to their needs. Which is a shame, since structured interpretation effectively guides the user through the entire process while providing unambiguous results—even across institutions. And although these developments open up opportunities for digital processing, free-form text remains a guaranteed option so as not to limit radiological expertise.