Awarded in 2026
Home Department: Radiology
Faculty Advisor: Craig Levin (Radiology)

Title: Interrogation of Two Markers of Neuroendocrine Tumors in a Single PET Scan to Guide More Precise Treatment

Abstract:
Neuroendocrine tumors are an uncommon cancer, and a single patient can have multiple tumor sites that behave very differently. Some sites carry a specific surface marker that doctors can image with a positron emission tomography (PET) scan using a tracer called 68Ga-DOTATATE; these sites can be treated with a targeted therapy that delivers radiation straight to them. Other sites have lost that marker and instead burn sugar at a high rate, so they appear on a different PET scan using a sugar-based tracer called 18F-FDG, and the targeted therapy does not work on them. To choose the right treatment, doctors need to see both behaviors, which means two separate scans on different days. That is hard on the patient, difficult to schedule, increases radiation exposure, and, because the body shifts between visits, makes it hard to tell whether a spot on one scan is the same spot on the other; as a result, two scans are never done in regular clinical practice. 

We are developing a way to capture both tracer images from a single scan. The two tracers emit signals differently: the sugar tracer releases a pair of photons, while the other releases that pair plus a third photon (a triple-photon signal). By having the scanner look for that extra photon, we can tell which signals came from which tracer and build a separate image for each. 

The main challenge is that the scanner sometimes mistakes stray photons for a real triple signal, which blurs the picture, and standard methods don't work well here. Our project has two parts. First, we will develop a computational method that uses timing and counting-rate information from the extra photon to estimate these false triples, aiming to keep the error below 3%, and validate it with realistic computer simulations. Second, we will test it on Stanford's new total-body PET scanner, first on patient-mimicking phantoms, then on prospectively collected scans from neuroendocrine tumor patients, and measure how the correction changes the balance between the two tracers at each tumor site, which doctors use to choose therapy. 

By combining physics, engineering, computing, and cancer medicine, this work aims to replace two PET scans with one, making both pieces of information practical to collect in the clinic and helping doctors choose the right treatment for patients whose needs are often overlooked.