Texas Health Resources

Jon Darnell

With over 20 years of experience in radiology and healthcare management, Jon Darnell is a strategic leader dedicated to advancing innovation in medical imaging, particularly through the integration of AI and technology advancements. Currently serving as the Radiology Director at Texas Health Harris Methodist Hospital Alliance, he oversees comprehensive radiology operations, driving substantial growth in imaging volumes and enhancing patient experience. Jon chairs the THR Radiology AI Steering Committee, guiding Imaging AI adoption to improve diagnostic accuracy and operational efficiency. His professional background includes notable contributions to healthcare publications, as well as expertise in budget management, cross-functional leadership and strategic planning for radiology infrastructure. Passionate about team development and excellence, Jon is committed to fostering a collaborative environment that supports patient safety, operational quality and innovative solutions in radiology.

Enhancing Imaging and Patient Care across Hospital Systems

In the last several years, my healthcare system has implemented several pieces of AI software in the radiology departments, both across the system and at individual hospitals. We have 13 wholly owned hospitals and six joint venture hospitals with three radiology physician groups reading across the enterprise. Recently, companies and vendors that develop and market AI for radiology have brought many, many software solutions to market. AI software for radiology has been developed for multiple modalities (i.e., CT, MRI and X-ray, for example) and is able to assess images from specific areas of the body to look for and identify disease states. This is to assist and enhance what a radiology physician would be identifying during their formal read of the study.

There are hundreds of vendors representing AI software and platforms for the radiology industry. Due to the recent explosion in this area of technology, a focused and structured approach to adopting this technology into a hospital practice is needed, whether it is for one hospital or a larger system.

Vendors are taking several approaches to market and implement this technology. The first approach is the vendor develops their own AI software. The vendor would market their software in individual software offerings or in package offerings that include several pieces of AI software, directly to end users/facilities. This means they employ or contract with all their own software developers. Most of their offerings are FDA approved, but some may not be. 

The second approach vendors use is licensing third-party developers of AI software to market to their customers. Again, this can be marketed as individual software offerings or as package offerings that would assess images of several different areas of the body. Some of these vendors may have their own platforms that could be launched directly from the end user’s current PACS solution.

The third approach vendors use for marketing AI for radiology applications is a hybrid model of the first two. These vendors develop their own software and license third party software to the market to end users/facilities. This allows the vendors to market software to end users that they would not otherwise have in their offerings.

"Knowing what specific AI software modules are strategically important to your hospital system imaging departments is key to determine what details to look for from AI software vendors"

There are two aspects radiology departments should consider when assessing AI vendors; how the vendor platform interacts with the radiology physicians during image assessment and how the AI platform real-time interfaces in order to communicate a positive finding to the care team. These two aspects are independent of each other and not all vendors will have either or both, of these platforms. If the vendor has an interactive platform for the radiology physicians for when they are reading the exams, the way that platform is utilized by the physician is important, as it can greatly affect the physician workflow efficiency. One example of an interface used during the exam reading task is the vendor AI software would send an assessed image with an interactive icon to the facility’s native PACS system. That icon on the image may indicate that the AI software has identified a finding and the physician could click on that icon inside the PACS system which would take the physician to the AI software, opening directly to the image where the positive finding can be found. Once the AI software has identified a positive finding, if the software platform has a results interface, it would then send the findings to the care team that is signed into the platform at the time. For instance, a positive finding for Large Vessel Occlusion (LVO) could be sent to the Neuro Interventionalist, Stroke/Vascular Cath Lab team, radiologist, stroke coordinator, Hospitalist and/or ED physician and any other clinical team member needing this information. Examples of software that a vendor may carry include intracranial, hemorrhage, LVL, brain aneurysm pneumothorax, aortic dissection and C spine fracture, just to name a few.  A stroke/neuro application that can detect a brain bleed, communicate that information to the provider and care team in a matter of seconds and allow an interventionalist to scroll through images on their mobile device to confirm a critical finding is just one way AI in imaging can and has, made a difference in patient care.  The seconds to get this information in today’s world vs. the minutes or hours that it once took to get communication for critical findings is the difference between life and death for many patients that benefit from today’s technology.

With so many aspects to consider, adoption of AI in medical imaging can be an overwhelming and arduous project, especially in a large hospital system. Knowing what specific AI software modules are strategically important to your hospital system imaging departments is key to determine what details to look for from AI software vendors. This would include pricing based on if the hospital would be purchasing one specific software module, a package of several software modules and if the vendor has a comprehensive platform and interfaces for radiology exam reading and the ability to communicate acute positive findings to the clinical care team.  Having a mobile application for fast communication to providers is essential to a Healthcare System to deliver time sensitive care to patients in dire situations, such as stroke and trauma.