A people-first approach guides technology transformation in radiology

In 2001, the BBC’s Jon Wurtzel introduced readers to a new piece of technology: a phone with a camera on it. “Infinite uses for the teenager, not entirely sure what the rest of us would do with one though,” said one reader. “Just another example of technological advances enticing us to pay ever more money for lower quality images,” said another. One forward-thinking reader commented that they would “use the camera phone to take pictures of my best friend: my dog Benson.” A phone prior to 2001 was just a phone, but it is hard to imagine one today without a camera. The difference is not just incremental improvement, but rather an understanding that true technological creativity is rooted in human needs and values. 

This philosophy is at the core of modern radiology practices. The primary goal of the field is to identify a disease, condition or injury based on its signs and symptoms. In this regard, the St. Jude Department of Radiology has leveraged a golden age of technology development to become a leading hub of imaging ingenuity. However, at the core of this tech boom is a very human commitment to improving the lives of patients that transcends any single technological advancement. 

VR brings everyone into the same space 

Virtual reality (VR) is emerging as a powerful tool for visualizing disease. By converting thousands of 2D scans into 3D virtual models, it is possible to preview a surgery — such as a tumor resection — from every angle, before ever needing to touch a scalpel. Zachary Abramson, DMD, MD, Department of Radiology assistant member, has championed this approach at St. Jude, where defining soft tissue boundaries between patient and tumor is both essential and extremely technical. This has been done in close collaboration with the Department of Surgery, particularly Andrew Davidoff, MD, Department of Surgery chair, who has been central to the endeavor. 

In a 2025 publication in Children, first and corresponding author Abramson and a multidisciplinary team outlined twelve specific guidelines to support image acquisition and 3D reconstruction of pediatric solid tumors. These reconstructions serve more than a clinical application; Abramson sees them as a powerful communications platform, bringing not just surgical teams, but also patients and families, into the conversation.

“We aim to provide surgeons with information they can’t get from 2D images alone. But over the past two years, we’ve also shared 3D models with more than 60 patients and families,” said Abramson. “We’ve expanded beyond surgeons using models in the intensive care unit, in palliative care and in cases involving chronic tumors or conditions, to get patients, family members and care teams all on the same page.”

In one case, the father of a 10-year-old child with an inoperable, but treatable tumor reached out to Abramson to help communicate the importance of his medication. “We met with him and his spouse, explained the anatomy to them and developed a model they could rotate and view on a phone,” Abramson said. “He’s going to use the model to have conversations with his child as they get older. That was extremely meaningful and not something I would have even thought about two or three years ago.”

Moving code to the clinic with Intelligent Imaging Informatics 

As virtual reality capabilities expand, accuracy remains critical; any deviations between model and tissue are not tolerated. This emphasis on safety runs deep through the department. In the realm of artificial intelligence (AI), radiology quickly became a case study, offering the chance to greatly expand imaging capabilities, measure tumors with a single click or automate reporting workflows. Yet, as swiftly as AI has broken through the cultural mainstream, concerns regarding the safety and reliability of both user inputs and generated outputs have emerged, tempering what initially appeared to be unchecked momentum. 

For Paul Yi, MD, Intelligent Imaging Informatics (i3) section chief and Department of Radiology associate member, this tempering is a necessary feature.

• A 2025 publication in Radiology from Yi assessed AI bias and provided best practices to ensure fair and equitable evaluation of AI systems for radiology.
• A 2024 publication in the Journal of Imaging Informatics in Medicine from Yi found that adult-trained deep-learning computed tomography (CT) organ segmentation models generalize poorly to pediatric patients, underscoring the importance of adapting AI tools intentionally for pediatric imaging.

“I’m interested in safe and trustworthy AI and how these technologies can be applied responsibly to unanswered clinical and biological questions,” said Yi. “This vision drives i3, the department’s hub for medical AI and informatics. By bringing together experts spanning computer science, biomedical informatics, information technologies and clinical radiology, we aim to bridge the gap between code and clinic.”

In practice, this is a deeply human effort. The comprehensive framework within i3 provides the oversight needed to ensure that any implementation of advanced technologies within radiology remains grounded in the steadfast commitment to improving patients’ lives.

• The Core Imaging Informatics (CI2) team manages the picture archiving and communication system behind all imaging operations and maintains the infrastructure for the other teams.
• The Tumor Metrics team provides consistent, reproducible tumor assessments to support clinical trials.
• The 3D Lab performs image segmentation and post-processing at the core of Abramson’s work, creating highly accurate 3D renderings and virtual reality models of tumors.
• The Image Quantification and Artificial Intelligence (IQAI) team combines AI development with data infrastructure to build and maintain the foundation needed for effective and responsible AI-informed imaging.

For Hailey Ross, MPH, i3 operations manager, this framework allows researchers and clinicians to glean as much information from their images as possible and think more clearly with data. “We have an incredible team of clinicians and researchers at St. Jude, but no one can be expected to know every tool available to them,” Ross said. “Through the systems we are putting in place, we can now help researchers recognize the value of their data and work together to uncover what’s hidden in the pixels.”

Radiopharmaceuticals ready to shine 

While technological breakthroughs inspire new ways of thinking, they also bring previously unattainable concepts within reach. One example is radiopharmaceuticals — molecules tagged with radiation-emitting atoms for clinical use — which sit at the intersection of chemistry, biology and physics. Decades of modernization across these disciplines have brought this field to a recent inflection point where safety profiles are unmatched, targeting is highly specific, use case is broad, and technology is primed. 

The process begins with the careful design of molecules that target a single location in the body, such as an infection or a specific disease marker. Kiel Neumann, PhD, Radiology Research section chief and Department of Radiology associate member, has developed several such probes:

• Published in Nature Biomedical Engineering, fluoroedaravone localizes to sites rich in reactive oxygen and nitrogen species, a hallmark of acute brain injury and neurodegenerative diseases such as Alzheimer’s disease.
• Published in the Journal of Nuclear Medicine, fluoromannitol specifically accumulates in infecting bacteria but not in mammalian host cells.

These molecules are synthesized with a radioisotope on one of the molecule’s atoms. For example, the radioisotope fluorine-18 is used to create 18F-fluoromannitol. This radioactive tracer emits signals detectible in tiny amounts using positron emission tomography (PET), where targeted sites in the body light up while surrounding areas remain unaffected.

With the framework for radiopharmaceutical research in place, Neumann is well positioned to help realize the potential of an exciting frontier in the field: theranostics. A combination of therapeutics and diagnostics, theranostics utilizes radiolabeled molecules in different ways to detect and treat diseases, especially cancers, with extreme accuracy and specificity. 

“The diagnostic side of theranostics exploits biomarkers unique to tumor cells to provide a comprehensive biological picture of disease, revealing metastases that might otherwise go undetected and informing clinicians on disease severity,” Neumann said. “The therapeutic side uses the same biomarkers to deliver radiation directly to cancer cells, causing DNA damage. We’re not targeting the proteins themselves; we’re using them like a Trojan horse to carry radiation into the tumor cell, as close to the nucleus as possible.”

In this way, theranostics can achieve a level of precision beyond that of even the most targeted chemotherapies. This will greatly benefit cancers for which therapies often have off-target effects, such as brain tumors. “We’re looking for ways to make the greatest impact on diseases with significant unmet needs; diffuse midline glioma is one major focus,” Neumann said. “We’re investigating biomarkers unique to these tumors and asking how they might be targeted with theranostics, and we’re encouraged by what we’re seeing.”

Keeping the human touch in radiology

As camera phones became the norm, software engineers and technology pioneers quickly found ways to integrate them into daily life while imagining what would come next. By staying true to their commitment to improve patients’ lives, researchers, clinicians and staff across the Department of Radiology are responsibly integrating next-generation technologies into clinical imaging and research while keeping an eager eye on unmet possibilities. The result is a future in which every scan reveals more, and every image has the potential to improve patient outcomes.