Researchers at Weill Cornell Medicine have received a five-year grant from the National Institute of Diabetes and Digestive and Kidney Diseases to support TRACE, a Tool for Reproducible, Accurate Contour Estimation. Driven by artificial intelligence, TRACE measures organ volumes from images of patients with autosomal dominant polycystic kidney disease (PKD). The standardized data and expert imaging analyses will be available to scientists nationally, which will help accelerate discoveries from basic research to clinical trials, ultimately improving care for people living with PKD.
This genetic condition causes many fluid-filled cysts to grow in the kidneys and interfere with function, eventually causing kidney failure in many cases. Approximately 600,000 people have PKD in the United States, and 90% of them have the autosomal dominant form of the disease, which is inherited when one affected parent passes on the DNA mutation to their offspring.
“The size of the kidney is currently one of the best ways of tracking disease progression and confirming that a treatment is working,” said principal investigator, Dr. Martin Prince, professor of radiology at Weill Cornell Medicine, and a radiologist at NewYork-Presbyterian/Weill Cornell Medical Center. “However, the traditional way of subjectively assessing the size of tissues and organs on images is being replaced by a new, deep-learning AI paradigm that enhances the accuracy of measurements, greatly reducing the variability of imaging data from one radiologist to the next.”
Dr. Mert Sabuncu, professor of electrical engineering in radiology at Weill Cornell, is also a grantee.
Building on decades of pioneering PKD imaging work by the team, the project aims to give researchers a “toolbox” for extracting precise imaging biomarkers—quantitative measurements that reveal what’s happening inside the body without invasive procedures. Physicians can use this information to choose the best treatment for a patient, such as medications to slow cyst growth or dialysis in the case of kidney failure.
Dr. Prince and his colleagues are developing a database of MRI and CT images of the kidneys and nearby organs, including the liver, pancreas, spleen and heart, from patients with PKD enrolled in research trials. To protect patient privacy, they have developed technology that strips all protected health information from images before they are uploaded to the database, the PKD Image Phenotyping Repository Core.
The database will include notes on organs showing their volume as well as the skeletal muscle index, which is a measure of muscle mass relative to body size, and the number and size of kidney cysts. This resource will help scientists measure disease progression and patient response to treatments over time. In addition, researchers will be able to study any organs within the field of view when the images were taken.
Researchers can explore available tools in a study setting.
“Our goal is to one day make the database freely available on the World Wide Web to give researchers access to the greater richness of information on these imaging studies, which will ultimately help to manage or possibly cure the disease,” Dr. Prince said.
