Verifiable and robust deformable image registration for precision tracking of diffuse glioma progression

The National Cancer Institute awarded Duke University $100,228 on December 1, 2026, under the Cancer Research Manpower program (CFDA 93.398) to develop a verifiable and accurate deformable image registration (DIR) method for tracking diffuse glioma progression and predicting recurrence sites in glioblastoma patients.

The project will create an AI-based image registration approach that incorporates blood vessel segmentation and bifurcation matching to achieve sub-millimeter accuracy in aligning longitudinal MRI scans. By establishing correspondence between blood vessels segmented from pre-operative and post-recurrence MRI scans, the method will synthesize vessel matching points with multi-sequence MRI image features to precisely describe anatomical transformation and pinpoint recurrence origin on preoperative imaging. This capability addresses a critical clinical gap: current imaging protocols lack precision to track microscopic tumor infiltration, resulting in poorly defined treatment margins that may not address the full tumor extent. The developed method is intended to enable refined treatment margins and support advanced imaging methods to accurately track tumor growth over time, potentially allowing boosted radiation doses to regions at risk for recurrence.

Work is performed at Duke University's Office of Research Administration in Durham, North Carolina, with a period of performance from December 1, 2026, through April 30, 2028. The Cancer Research Manpower program supports biomedical training opportunities for individuals pursuing careers in basic, clinical, and prevention research relevant to the National Cancer Program.

Abstract

Diffuse gliomas are the most common primary brain cancer in adults, with the most aggressive and common form glioblastoma (GBM) having a median survival of only 15 months. Despite advances, current clinical imaging protocols lack the precision to track microscopic tumor infiltration that later becomes recurrence. This can result in poorly defined treatment margins that do not address the full extent of the tumor and can compromise patient safety. Determining the precise site of future recurrence on preoperative imaging could allow improved treatments such as boosted radiation dose to regions at risk for recurrence. Deformable image registration (DIR) enables the alignment of longitudinal MRI scans to achieve this task, but existing DIR approaches suffer from limited accuracy and unreliable verification, hindering clinical adoption. This project aims to develop a verifiable and accurate DIR method for diffuse gliomas by incorporating AI-based blood vessel segmentation and bifurcation matching into a new high-precision DIR approach. I hypothesize that incorporating blood vessel bifurcations as stable anatomical markers into the registration process will permit sub-millimeter accuracy in non-linear image registration. This level of accuracy will permit refined and precise treatment margins and support advanced imaging methods to accurately track tumor growth over time. Using the hierarchical nature of blood vessel trees, I will establish correspondence between blood vessels segmented from pre-operative and post-recurrence MRI scans of the same GBM patient. The DIR method will synthesize these matching vessel points with image features extracted from multi-sequence MRI data to precisely describe the anatomical transformation that occurred between the scans. This can then be used to pinpoint the recurrence origin on the pre-operative time point, allowing for future treatments targeting this site. In addition to GBM, I will adapt and optimize the developed method for slower-growing, low-grade diffuse glioma cases. By quantifying the deformation between low-grade scans with accurate DIR, I can detect small changes in tumor size and shape that can indicate disease progression. I will compare this approach to visual observation to demonstrate its clinical utility. Finally, I will utilize the developed vessel-matching tools to establish the most comprehensive DIR accuracy baseline across diffuse glioma grades to date, supporting further algorithm development and clinical implementation. This research will provide robust, verifiable DIR methods tailored for GBM and low-grade diffuse gliomas, addressing a critical GAP in neuro-oncology imaging. By improving registration accuracy, this project can improve the precision of GBM treatment planning, enhance recurrence detection, detect tumor progression, and ultimately improve patient outcomes.