The logic behind the conventional approach is simple: because glioblastoma cells creep microscopically into surrounding tissue, the radiation field must be larger than the visible tumor. But histopathologic and preclinical studies have long shown that glioblastoma does not spread isotropically. Instead, it migrates preferentially along large white matter pathways, the bundled fiber tracts that connect distant brain regions. The tumor behaves less like an inflating balloon and more like water flowing through a network of pipes, racing along highways of myelinated axons while slowing at natural barriers. Yet this biological reality has never been systematically built into radiotherapy planning, until now.

A team led by Michael Wahl of the University of California, San Francisco, together with colleagues across UCSF’s radiation oncology, neurology, radiology and neurosurgery departments, has now translated that insight into a clinically workable technique. Their method uses diffusion-weighted MRI, which measures how freely water molecules diffuse through tissue. In organized white matter, water diffuses preferentially along the axis of the fiber bundles, a phenomenon called anisotropy. Tractography algorithms exploit this signal to reconstruct the trajectories of white matter fiber pathways throughout the entire brain, effectively drawing a wiring diagram of each individual patient.

The technical pipeline behind the study is a careful exercise in modern imaging science. Thirteen glioblastoma patients underwent 55-direction high angular resolution diffusion imaging on a 3 Tesla scanner as part of their postoperative radiation planning MRI, an acquisition of only seven to eight minutes. After correcting the data for head motion and eddy-current distortion using the FMRIB Software Library, the researchers fit both a tensor model and a higher-order Q-ball model with the open-source DIPY package, the latter to accurately resolve regions where fiber bundles cross. White matter, defined as voxels with fractional anisotropy above 0.15, was seeded densely, and a residual-bootstrap tractography approach estimated fiber orientations probabilistically, terminating streams when anisotropy fell below threshold or fibers bent more than sixty degrees.

Here is where the innovation crystallizes. The team selected all tractography streamlines passing within five millimeters of the visible tumor, the gross tumor volume delineated from the surgical cavity, contrast-enhancing regions and mass-like T2-FLAIR abnormality. From this subset they generated a scalar map in which every voxel stores its minimum white matter path length to the tumor, the shortest distance along actual fiber tracts rather than straight-line distance. The result is an anatomically informed expansion: the target flows along the superior cingulum bundle in one patient, crosses to the opposite hemisphere through the splenium of the corpus callosum in another, and in a third tracks along the inferior longitudinal fasciculus, precisely the routes along which those tumors later spread. In every one of the thirteen cases, the resulting contours respected natural anatomic boundaries that the standard circular expansion ignored.

The payoff appears in the numbers. When the researchers generated clinical target volumes by thresholding the path length maps at two centimeters and compared them, within the same patient, to the guideline-conformant two-centimeter isotropic expansion, the tractography-based volumes were a median of 67 cubic centimeters smaller, roughly a nineteen percent reduction, a difference that reached statistical significance. The one-centimeter variant shrank targets even more dramatically, by 39 percent on average, though at the cost of missing several recurrences captured by the standard volume. At three centimeters, the tractography volume converged with the standard one, as expected. These volumes were imported directly into the RayStation treatment planning system, demonstrating that the approach is compatible with routine clinical workflows rather than confined to research software.

The more consequential question is whether these sculpted volumes actually contain the places where tumors come back. To answer it, the team performed a patterns-of-failure analysis, rigidly registering each patient’s recurrence MRI to the planning scans and classifying recurrences as central, in-field, marginal or distant relative to each target definition. The median time to recurrence was twelve months. Ten of the thirteen recurrence volumes fell entirely within the conventional isotropic target, but twelve of thirteen were encompassed by the two-centimeter tractography-based volume, all while that volume was substantially smaller. In two cases, the recurrent tumor was clearly infiltrating along a major white matter tract inside the tractography target but outside the standard one, exactly the failure mode the technique was designed to prevent.

The clinical context makes these findings timely. Historical teaching held that ninety percent of glioblastoma recurrences occur within a two-centimeter margin of the original tumor, a figure that discouraged tinkering with target delineation. But more recent studies of patients treated with concurrent temozolomide, particularly those whose tumors carry methylation of the MGMT promoter and are therefore more drug-sensitive, have shown marginal or distant recurrence rates approaching forty percent, suggesting that current volumes may undertreat pockets of microscopic disease in some patients. At the same time, generous circular expansions irradiate substantial volumes of healthy brain, raising the specter of radiation necrosis and long-term neurocognitive decline. An approach that could simultaneously tighten targets and extend them along true routes of invasion promises the rarest of things in oncology: better coverage with less collateral damage.

The researchers are appropriately measured about what their study can and cannot claim. Thirteen patients cannot prove a statistical reduction in marginal or distant recurrences, and the pattern-of-failure analysis was observational and hypothesis-generating. Even demonstrating that a recurrence fell within a tractography target does not prove that treating that volume would have prevented it, since most patients still recur within the high-dose region regardless of technique. The small sample also prevented confident optimization of the tracking parameters and path length thresholds, and no dosimetric analysis of organ-at-risk sparing has yet been performed. Validation in a larger prospective cohort, with formal assessment of normal brain sparing through dose-volume statistics, is the clearly stated next step.

Still, as a proof of concept, the study is striking in its feasibility and simplicity. The imaging sequence adds only minutes to a standard planning MRI, the analysis relies on open-source tools, and the output drops directly into commercial treatment planning software. What the UCSF team has shown is that the brain’s own anatomy can be enlisted as a map of where an incurable tumor is likely to travel next, replacing a geometric convention inherited from decades-old CT studies with a personalized, biologically grounded contour. If larger trials confirm that two centimeters along a nerve fiber tract is worth more than two centimeters in a straight line, the humble circle that has defined glioblastoma radiotherapy for a generation may finally be retired, and one of medicine’s grimmest diagnoses may gain a small but meaningful edge.

Subject of Research: Tractography-based clinical target volume delineation for glioblastoma radiotherapy planning

Article Title: White matter pathlength maps from diffusion-weighted MRI tractography for radiotherapy target planning in glioblastoma

Article References: Wahl, M., Chapman, C. H., Morin, O., Jordan, K., Henry, R. G., Chang, S. M., Villanueva-Meyer, J. E., Mukherjee, P., Theodosopoulos, P., McDermott, M. W., Berger, M. S., Sneed, P., Braunstein, S. E., & Lupo, J. M. (2026). White matter pathlength maps from diffusion-weighted MRI tractography for radiotherapy target planning in glioblastoma. Journal of Neuro-Oncology, 179(3), Article 102. https://doi.org/10.1007/s11060-026-05789-9

Image Credits: AI Generated

DOI: 10.1007/s11060-026-05789-9

Keywords: glioblastoma, diffusion-weighted MRI, tractography, radiotherapy planning, white matter, clinical target volume, patterns of failure, tumor recurrence, neuro-oncology, UCSF, temozolomide, medical imaging

The clinical logic behind the work rests on a well-known but stubbornly difficult fact about glioblastoma biology: the disease rarely returns in a random location. Most patients experience recurrence at or close to the cavity left behind after the tumor is removed. That means the seeds of the next tumor are almost certainly already present in the tissue surrounding the resection site at the time of surgery, invisible to conventional imaging and often to the surgeon’s eye as well. If clinicians could identify which stretches of that surrounding tissue harbor infiltrating tumor cells, they could in principle act on that knowledge immediately, either by removing additional tissue during the initial operation or by directing focused therapies at the highest-risk zones before the disease re-establishes itself.

Turning that idea into a practical tool required solving a fundamental bottleneck in operating room pathology. The research team, led by first author Sanjeev Herr, MD, a postdoctoral research fellow at UC San Francisco, analyzed tissue samples collected during glioblastoma surgery from patients treated at UCSF Health, for whom the median time to recurrence was 5.5 months. Rather than relying on conventional pathology, which requires tissue to be fixed, processed with dyes and stains, and interpreted by a pathologist in a process that is both slow and labor-intensive, the team used a technique called stimulated Raman histology, or SRH. SRH produces microscopic images of fresh, unprocessed tissue in less than a minute, preserving the tissue in a near-native state and dramatically compressing the timeline from sampling to analysis.

Stimulated Raman histology works by exploiting the intrinsic vibrational signatures of the molecules inside tissue. When laser light interacts with lipids, proteins, and other cellular components, the scattered light carries chemical information that can be translated into images resembling traditional stained histology, all without a single drop of dye. That speed and chemical richness make SRH an ideal input for machine learning, because the raw optical data can be fed directly into an algorithm trained to recognize patterns that correlate with tumor infiltration. In this study, the researchers paired SRH imaging with FastGlioma, an AI system previously developed by investigators at UCSF and the University of Michigan, which scores tissue based on the degree of tumor infiltration it detects.

The scale of the study reflected the rigor needed to validate such a tool. The researchers developed their model using approximately 300 tissue samples drawn from 60 patients, then tested it separately on about 100 samples from a different set of 20 patients, keeping the training and testing data strictly apart. The central question was whether the AI’s assessment of infiltration, computed from fresh tissue in minutes, could predict where recurrent tumor would later emerge. The answer was encouraging on its own: the AI infiltration score alone performed about as well as conventional pathology at predicting which sampled areas would later develop recurrent disease. In other words, the algorithm extracted essentially the same prognostic signal from unprocessed tissue that the gold standard of surgical pathology extracts from stained slides, but in a fraction of the time.

The most striking results, however, came when the team layered additional information on top of the AI score. The researchers combined the infiltration measure with clinical, imaging, and molecular data and tested the resulting feature set across six different machine-learning models. The best-performing model was significantly more likely to distinguish between tissue sites that would and would not go on to develop recurrences than any single data type could manage alone. Notably, the AI measure of tumor infiltration proved to be the strongest individual predictor of recurrence in five of the six models, providing more predictive information than the tumor’s molecular characteristics, which are traditionally among the most powerful prognostic markers in glioblastoma. That finding suggests the spatial pattern of microscopic invasion, as read by the algorithm, captures something about recurrence risk that even genomic profiling misses.

Precision mattered as much as predictive power. A tool that says recurrence is coming somewhere in the general vicinity of the tumor cavity would be of limited use to a surgeon deciding where to extend a resection or where to aim a radiation boost. The researchers therefore tested how precisely the model could localize the future site of relapse, and it performed well at predicting whether cancer would return within 5 or 10 millimeters of the tissue that had been sampled. At those spatial scales, the difference between a predicted high-risk zone and an unremarkable one becomes actionable: five to ten millimeters is within the range a neurosurgeon can meaningfully consider when weighing how much additional tissue to remove, and it is squarely within the targeting capability of modern focal radiation techniques.

The therapeutic implications flow directly from that precision. Senior author Shawn Hervey-Jumper, MD, a neurosurgeon at UCSF Health and the Mitchel S. Berger, MD, endowed professor at the Weill Institute for Neurosciences, outlined two distinct pathways by which the predictions could change practice. For patients whose high-risk tissue lies in regions of the brain that can safely be resected, the AI guidance could support removing additional tissue during the initial surgery, clearing infiltrating cells before they have a chance to repopulate the cavity. For patients whose disease sits in parts of the brain that cannot be removed, the predicted sites of progression could instead be targeted with other treatments, such as higher-dose focal radiation or drugs infused directly into the tumor through a catheter placed through the skull. In both scenarios, the goal is the same: intervene while the residual disease is still microscopic and most vulnerable.

The system’s developers also see it functioning as a real-time decision support tool rather than a post-hoc analysis. Herr described the platform as having incredible potential to provide neurosurgeons with valuable real-time guidance during tumor removal, while also generating insights that guide subsequent treatment decisions. Because SRH imaging and AI scoring can be completed in under a minute on fresh tissue, the workflow could in principle fit inside a single operation, giving the surgical team infiltration maps while the patient is still on the table. That stands in sharp contrast to the current standard, in which definitive pathology results often arrive days after the operation, when the opportunity to act on them surgically has already passed.

The researchers deliberately focused their predictions on the first recurrence, a choice that reflects both scientific caution and clinical strategy. Patients with glioblastoma often receive experimental treatments later in the course of their disease, and those interventions can influence tumor growth in ways that make subsequent recurrences far more difficult to predict reliably. The first recurrence, arriving before most of those confounding therapies, offers the cleanest signal for validating a predictive model. It is also the recurrence that matters most strategically, as co-senior author Todd Hollon, MD, of the Machine Learning in Neurosurgery Laboratory at the University of Michigan, Ann Arbor, emphasized: the overall goal was to delay that first recurrence, and ultimately the team hopes that extending that window could translate into longer survival. For a disease in which median survival is measured in months and every extension of progression-free time carries real weight, a tool that converts fresh tissue, rapid optical imaging, and machine learning into a map of where the tumor will strike next represents one of the more concrete near-term applications of AI in neuro-oncology, and a potential shift from reacting to recurrence toward anticipating it.

Subject of Research: AI prediction of glioblastoma recurrence sites using stimulated Raman histology

Article Title: Treating a deadly brain tumor before it recurs? AI tool may help

Article References: Treating a deadly brain tumor before it recurs? AI tool may help. (n.d.). Original publication

Image Credits: AI Generated

DOI: Not provided

Keywords: glioblastoma, artificial intelligence, stimulated Raman histology, brain tumor recurrence, FastGlioma, neurosurgery, machine learning, tumor infiltration, focal radiation, UCSF, Science Advances, precision medicine