A multidisciplinary team of Vanderbilt researchers has received a prestigious R37 grant from the National Cancer Institute (NCI), part of the National Institutes of Health (NIH), to develop an enhanced visualization and navigation system for the endoscopic treatment of upper tract urothelial carcinoma (UTUC), a cancer of the kidney’s inner collecting system. The award will provide approximately $2.8 million over five years, with the possibility of a two-year extension.
The project, “Improved Surgical Visibility and Navigation during Endoscopic Treatment of Upper Tract Urothelial Carcinoma,” brings together Jie Ying Wu and Ipek Oguz, both computer science researchers and affiliates of the Vanderbilt Institute for Surgery and Engineering (VISE), with Dr. Nick Kavoussi, a urologist at Vanderbilt University Medical Center (VUMC).
UTUC is diagnosed in roughly 20,000 patients in the United States each year. Of those, 30% will experience a recurrence after initial treatment, and 20 to 30% will eventually develop metastatic disease. The current gold-standard treatment, complete removal of the kidney and ureter, is effective but comes with a significant cost: decreased kidney function is linked to increased illness and death. A minimally invasive, kidney-sparing alternative exists, but it is technically difficult and has been slow to catch on.
“The goal of endoscopic surgery is to completely visualize the intrarenal collecting system to identify and examine all tumors while keeping track of their locations during surgery,” Kavoussi said. That depends on a surgeon’s ability to mentally reconstruct a 3D map of the kidney’s branched internal anatomy from 2D imaging captured before surgery, a task Kavoussi described as extremely difficult. Because tumors can appear anywhere within the kidney’s networked calyces, tracking what has and has not been examined creates a heavy cognitive burden during surgery. As a result, patients often require multiple operations to fully localize and treat their tumors, and many end up undergoing more invasive, tissue-removing surgery just to ensure the cancer is completely treated.
The Vanderbilt team’s project aims to change that by combining computer vision and 3D reconstruction models to give surgeons real-time guidance during the procedure. Wu, whose lab focuses on computer vision, machine learning and augmented reality for surgery, models the surgical scene by pulling together information from multiple sensors throughout an operation. That modeling can be used to feed surgeons additional information through augmented reality or, eventually, to support robotic surgical automation.
Oguz’s lab brings a complementary expertise in medical image computing, working across imaging modalities such as MRI and ultrasound and organs ranging from the retina to the placenta to the brain, with algorithms designed to extract subtle patterns that aren’t visible to the naked eye. “Specifically in this project, we will be working with both preoperative CT images and the endoscopic video during the surgery to empower the surgeon to make better informed decisions,” Oguz said. That means helping with surgical navigation so surgeons get a clearer endoscopic view of the anatomy during the procedure, and also helping interpret the video feed in real time to support treatment decisions as they happen.
“Traditional 3D reconstruction techniques work great for scenes with a lot of structure and texture, like buildings, but in human anatomy, the texture is much more subtle,” Wu said. “AI tools can be trained to recognize this subtle texture from data and create reconstructions in this challenging environment.” Wu added that AI has also accelerated the underlying processing enough that reconstructions can guide surgeons in real time rather than after the fact.
AI is central to Oguz’s lab as well, she said, describing her team as both developers and consumers of these technologies. “Depending on the problem at hand, we might develop something from scratch, or perhaps we might look at the broader field of computer vision,” Oguz said. “Maybe somebody has developed a relevant method but in the context of astronomy images or analyzing videos of soccer games in the World Cup, and we then think about how to adapt that to endoscopic kidney videos, which has unique challenges.”
Kavoussi said the team plans to use computer vision models to help localize and characterize tumors within the kidney’s collecting system, including grading information that can inform treatment decisions, alongside reconstruction models that map the kidney for real-time guidance and help ensure the entire organ is fully evaluated.
The team hopes the resulting system will make it possible for more surgeons to offer minimally invasive treatment for UTUC, sparing patients more invasive procedures and preserving kidney function. “The hope is that this work allows clinicians to make more confident and accurate decisions in real time during endoscopic treatment of UTUC,” Kavoussi said. “This would save many patients from undergoing a more invasive surgery when they can be treated in a minimally invasive manner.” Wu said the tool could also ease the mental workload for surgeons who already perform the procedure, and eventually be integrated as a software addition to existing endoscopic surgical systems, opening the door to large-scale clinical deployment.
Oguz pointed to a broader benefit as well: helping hospitals that don’t have access to the kind of specialized, high-volume teams Vanderbilt can offer. “At Vanderbilt we are fortunate to have teams of experts in many clinical specialties, including this one,” she said. “But if you imagine a rural hospital, they may have fewer opportunities to build experience with these complex procedures, as they may have lower case volume and limited access to specialized support. We are hoping that the computational tools we develop would help them provide higher quality care to their patients.”
For Wu, the recognition carries added weight because of the years-long collaboration behind it. “Having NIH recognize Dr. Kavoussi’s and the whole team’s work to advance minimally invasive kidney treatments is very exciting,” she said. “To me, it is an indication of the potential for this work to really improve surgical care for patients and the value of interdisciplinary collaboration.” Kavoussi echoed that sentiment: “We are honored by this recognition, as it acknowledges the potential of a strong, multidisciplinary collaboration to transform patient care.” Oguz agreed, noting the award caps off years of groundwork. “We are super excited, as this is the result of several years of hard work to build this collaboration from the ground up,” she said. “We especially think the five-plus-two-year format will give us the necessary room to really dig deep and come up with innovative ideas to tackle these technical challenges.”
Both researchers point to VISE as the reason the collaboration exists in the first place. Wu and Kavoussi met through the institute’s network and have worked together for several years, with early-stage projects supported by VISE seed funding. “VISE has provided the network, the physical space and seed funding to bring this team together,” Wu said. She noted that VISE’s location inside the hospital has been essential for running user studies with surgeons, and that its proximity to VUMC has already allowed the team to deploy and test a related guidance system, developed with Kavoussi, in nearly 100 patient cases. Kavoussi called VISE “a venue for like-minded, impassioned collaborators to think of creative solutions to complex clinical problems,” adding that the R37 proposal “demonstrates the benefit of an entity like VISE in facilitating translational research.”
Oguz, who also met Kavoussi through VISE, described the project as emblematic of what the institute makes possible. “This project must be the poster child for VISE,” she said. “Nick and I have had pretty much every kind of support: VISE supported summer undergraduate students in the earliest days, we received a VISE physician-in-residence award for Nick to help him protect research time, and one of my Ph.D. students was supported by the T32 program.” Beyond those specific contributions, Oguz said her lab benefits daily from the intellectual and physical environment VISE has built. “This grant is the culmination of all those investments coming together,” she said.
The project also reflects the College of Connected Computing’s emphasis on interdisciplinary research and human-centered design. Wu said her work involves close, regular collaboration with clinicians at VUMC, including brainstorming sessions to sketch out how technology might address specific clinical challenges. Before writing any code, her team runs focus groups to understand a clinical problem from multiple perspectives and to co-design potential solutions with the surgeons and staff who will eventually use them. Prototypes are then tested in phantom models with surgeons and trainees, and once a design is finalized, the team can test it in the operating room just steps from VISE’s space in the hospital. “This frequent iteration enables us to carry out human-centered design and ensure the technology we develop actually addresses a clinical need,” Wu said.
Oguz described the same principle at work in her own lab. “This project is inherently interdisciplinary, bringing together a team of surgeons and computer scientists,” she said. “If we developed these technologies in isolation in my lab, we could easily create methods that are technically sophisticated but don’t address the realities of the operating room. Instead, we are collaborating very closely with Nick and his team to make sure we are very tightly aligned every step of the way. This allows us to create solutions that fit naturally into the surgical workflow and have real potential to benefit surgeons and patients.”
The R37 award formally launches a project years in the making, one the team hopes will ultimately reshape how upper tract urothelial carcinoma is treated, giving more patients access to a kidney-sparing option and more surgeons the tools and confidence to offer it.