The core problem with conventional smoothing lies in its indifference to brain anatomy. A Gaussian kernel treats the brain as a continuous 3D volume, aggregating signal from any voxel within its radius regardless of whether those voxels belong to the same gyrus, the same cortical layer, or even the same tissue type. When the cortex folds in complex sulcal patterns, two voxels that sit just millimeters apart in Euclidean space may be separated by a deep sulcus and belong to functionally distinct regions. Smoothing across that boundary blurs signal from one gyrus into its neighbor, creating activation clusters that never existed biologically. The white-gray matter interface presents an even more vulnerable boundary, as signal from cortical gray matter can leak into white matter and artificially shift the apparent peak of activation away from the cortex.

Surface-based fMRI analysis has long been considered the gold standard for avoiding these volumetric smoothing artifacts. By projecting data onto an unfolded cortical surface, this approach inherently respects the geodesic distance along the cortex and prevents signal from crossing sulci. But surface-based workflows require complex preprocessing pipelines, specialized file formats like CIFTI, and a small set of tools that may not integrate with researchers’ preferred software. Many established neuroimaging packages still lack surface-based capabilities, leaving most labs to choose between adopting an entirely new analytical paradigm or continuing with volumetric smoothing and accepting its artifacts.

Ellis and Aizenberg built their constrained smoothing tool as a practical middle ground. The method represents each brain voxel as a node in a graph, with edges connecting all 26 neighboring voxels in 3D space. Critically, the researchers then prune any edge that crosses either the pial or white matter surface as reconstructed by FreeSurfer, using ray tracing to rapidly evaluate surface intersections across the more than 40 million nodes in a typical brain graph. After pruning, the graph separates into four independently smoothable components—left and right gray matter and left and right white matter—ensuring that smoothing kernels can never aggregate signal across the cortical boundaries. The researchers implemented the approach in Python, avoiding the licensing costs of MATLAB, and optimized the heat kernel smoothing parameter to approximate a user-specified Gaussian full-width at half-maximum, allowing direct comparison between constrained and unconstrained smoothing at equivalent kernel widths.

To evaluate the method, the researchers applied both smoothing approaches to simulated data, a sensory task fMRI dataset, a precision motor mapping dataset from the Midnight Scan Club, and resting state fMRI data from the Human Connectome Project in Aging. In the simulation experiments, where 68 cortical regions were isolated and white noise added, constrained smoothing accurately recovered the shape of the original regions while Gaussian smoothing created artifacts by smoothing across boundaries. Constrained smoothing demonstrated higher accuracy in terms of both Dice similarity coefficient and Pearson correlation across all tested kernel widths, and more accurately concentrated positive predictions within gray matter while reducing white matter activation.

The real-world task fMRI results told a similarly compelling story. In the sensory task analysis, where participants received brushing stimulation to the left hand, both Gaussian and constrained smoothing increased the reliability of activations across repeated acquisitions. However, Gaussian smoothing significantly increased the percentage of active voxels in white matter relative to constrained smoothing. In the motor task experiments, where each of ten subjects completed 20 runs of finger tapping, toe tapping, and tongue movement, Gaussian smoothing with kernel widths greater than 3 mm decreased the accuracy of individual session maps compared to constrained smoothing. Notably, smoothing did not uniformly improve accuracy across all metrics—Dice scores were actually highest with no smoothing at all, suggesting that the reliability gains from smoothing come at a cost to spatial precision.

The false positive analysis addressed a critical statistical concern. Because constrained smoothing is inherently anisotropic, it produces smoothed data that may violate the Gaussian random field assumptions underlying parametric statistical models in packages like FSL FEAT. Previous research by Eklund and colleagues has shown that such violations can lead to inflated false positive rates in fMRI studies. Ellis and Aizenberg tested this by running task fMRI analysis on resting state data—data acquired with no task being performed—so that any detected activation could be considered a false positive. With cluster significance thresholding, mean false positive voxel percentages remained below 5% for both smoothing methods across all tested kernel widths, providing reassurance that constrained smoothing does not dramatically inflate statistical errors when appropriate thresholding is applied.

Perhaps the most surprising findings emerged from the resting state connectivity analysis. Using the Schaefer/Yeo 1,000-parcel atlas, the researchers fit generalized additive mixed models to examine how smoothing altered the relationship between Euclidean distance and functional connectivity. Both Gaussian and constrained smoothing inflated connectivity between spatially proximal regions, but Gaussian smoothing demonstrated more inflation at kernel widths of 6 mm or greater. Even more striking, both smoothing methods significantly altered graph theory metrics including clustering coefficient, global efficiency, local efficiency, and modularity under all tested kernel widths. Constrained smoothing had less effect on local efficiency, global efficiency, and clustering coefficient at higher kernel widths, though the differences between methods were marginal compared to the overall effect of smoothing itself on these network measures.

The researchers acknowledge an important caveat: constrained smoothing is inherently more conservative than unconstrained smoothing at identical nominal kernel widths. By restricting connections across cortical boundaries, constrained smoothing aggregates fewer data points, effectively lowering the dimensionality of the smoothing operation. The effective level of smoothing also depends on graph connectivity, which in turn depends on voxel size—smaller voxels produce more connections and more data aggregation. This suggests constrained smoothing may be most advantageous in high-resolution fMRI experiments, where the additional anatomical detail justifies the computational cost of graph construction and pruning.

As a practical alternative to both unconstrained volumetric smoothing and full surface-based analysis, anatomically constrained volumetric smoothing offers researchers a way to enhance fMRI reliability while keeping data in the familiar voxel space. The method requires reconstructed cortical surfaces but avoids the need to transform data into surface-based representations, preserving compatibility with decades of established volumetric processing pipelines. Future work could extend this graph-based approach to smooth along white matter tracts, integrate with Bayesian inference frameworks that determine optimal smoothing from the data, or apply nonparametric permutation testing for valid statistical inference on constrained-smoothed volumetric data. For now, the study provides compelling evidence that the smoothing step so many researchers take for granted deserves a second look—and that respecting the brain’s own boundaries may be the key to more accurate brain maps.

Subject of Research: Anatomically constrained volumetric smoothing enhances fMRI reliability while avoiding smoothing artifacts

Article Title: Anatomically constrained volumetric smoothing enhances fMRI reliability while avoiding smoothing artifacts

Article References: Ellis, D. G., & Aizenberg, M. R. (2026). Anatomically constrained volumetric smoothing enhances fMRI reliability while avoiding smoothing artifacts. Frontiers in Neuroimaging, 5, Article 1753534. https://doi.org/10.3389/fnimg.2026.1753534

Image Credits: AI Generated

DOI: 10.3389/fnimg.2026.1753534

Keywords: Anatomically, constrained, volumetric, smoothing, enhances, fMRI, reliability, while, avoiding, artifacts, scientific research

The retrospective analysis drew on an institutional database of patients who had received first-time stereotactic radiosurgery for a single brain metastasis and subsequently underwent surgical resection of that same lesion, providing the gold standard of histopathological confirmation for what the imaging had shown. Thirty-six patients met the inclusion criteria. For each patient, the team manually measured tumor volumes before treatment and at three standardized post-treatment intervals: six weeks, three months, and six months. Rather than relying on a single measurement dimension, the investigators segmented three distinct tissue compartments on each scan: the contrast-enhancing volume, the FLAIR hyperintense volume, and the necrotic component. From these they derived two ratios intended to capture the relationship between edema and tumor burden, the FLAIR-to-enhancing ratio and the enhancing-to-necrotic ratio, and then modeled how each parameter evolved over time.

The volumetric data revealed a treatment effect that was real but incomplete. Mean enhancing volume declined by 28.5 percent from baseline at six weeks, 25.8 percent at three months, and 18.9 percent at six months, a pattern consistent with an initial radiosurgical response that partially attenuated over the follow-up window. The FLAIR volume followed a different and more revealing course: it fell by 16.9 percent at six weeks and 22.2 percent at three months, but then rebounded to nearly baseline, changing only 2.4 percent from the pretreatment measurement at six months. This divergence between the enhancing core and the surrounding hyperintensity is precisely why the authors argue that single-compartment assessment, the standard practice in most surveillance protocols, misses a substantial portion of the biological story unfolding after radiation.

Statistical analysis confirmed that the two compartments do not move independently. Percent changes in FLAIR volume and enhancing volume correlated significantly at both the six-week time point, with a p-value of 0.004, and the three-month time point, where the association was even stronger at p less than 0.001. In practical terms, lesions whose enhancing cores were shrinking tended to show shrinking edema as well, at least in the early post-treatment period. But the correlation broke down in the cohort’s most clinically important subgroup: the patients whose lesions ultimately failed treatment and required resection. When the investigators stratified the surgical outcomes by pathology, the FLAIR-to-enhancing ratio trended differently between radiation necrosis and recurrent tumor, and patients who developed radiation necrosis had significantly lower FLAIR volumes at six weeks compared with those whose lesions harbored recurrent tumor.

This finding carries a counterintuitive implication. Clinicians often assume that more edema signals more trouble, and in the context of symptomatic mass effect that intuition holds. But the data suggest the opposite relationship for diagnostic purposes: a lesion that retains relatively little FLAIR hyperintensity six weeks after radiosurgery, despite failing to shrink, is more likely to be radiation necrosis, whereas a non-responding lesion wrapped in substantial and persistent edema raises greater concern for viable, actively growing tumor. The biological rationale is plausible. Radiation necrosis is fundamentally a vascular and inflammatory process driven by endothelial injury, blood-brain barrier disruption, and cytokine-mediated swelling, whereas recurrent tumor generates edema through tumor-cell infiltration, angiogenic signaling, and active secretion of vascular permeability factors. The two processes stamp different volumetric signatures on the surrounding brain, and the early post-treatment window appears to be when those signatures are most distinguishable.

Beyond the diagnostic ratios, the study identified which patient and lesion characteristics shape the edema trajectory itself. Linear mixed-effects models, a statistical framework well suited to repeated measurements nested within individual patients, showed that baseline lesion size, baseline FLAIR volume, and the FLAIR-to-enhancing ratio all influenced FLAIR trajectories with p-values below 0.001. Primary cancer type and sex also emerged as significant factors, each with p-values of 0.02. These results echo a growing body of literature indicating that peritumoral edema after radiosurgery is not a uniform phenomenon but one modulated by tumor histology, systemic therapy exposure, and intrinsic lesion characteristics. Larger lesions and those with greater baseline edema burdens follow different swelling curves than small, dry lesions, which means that any future edema-based diagnostic criterion will likely need to be normalized to baseline values rather than applied as an absolute threshold.

The clinical context makes the stakes of this work concrete. Stereotactic radiosurgery has become a cornerstone of brain metastasis management precisely because it delivers ablative doses of radiation with submillimeter accuracy while sparing surrounding tissue, allowing many patients to avoid whole-brain radiation and its cognitive sequelae. But local failure rates remain nontrivial, and a meaningful fraction of patients with non-responding lesions eventually come to surgery. Prior biopsy series have shown that enlarging lesions after radiosurgery frequently turn out to be radiation necrosis rather than tumor, which means that a substantial number of patients may currently undergo resection, or receive escalating systemic therapy, for a process that was never malignant. Conversely, misattributing recurrent tumor to necrosis delays effective treatment in a population where every month of uncontrolled intracranial disease matters. Any imaging tool that sharpens this fork in the road has immediate utility.

The Yale team’s methodological choices deserve attention because they point toward how such a tool could be deployed at scale. Manual segmentation of FLAIR hyperintensity is labor-intensive and subject to interobserver variability, which has historically limited edema volumetry to research settings. The authors explicitly recommend that automated edema volumetry be incorporated into routine post-SRS surveillance, and their proposal aligns with recent pilot work demonstrating that picture archiving and communication system-integrated volumetric tools can provide peritumoral edema measurements that add information beyond standard response assessment criteria for brain metastases. If FLAIR volume trajectories can be generated automatically at each surveillance scan, the six-week and three-month ratios identified in this study could become practical decision-support metrics, flagging non-responding lesions whose edema behavior suggests recurrent tumor for earlier tissue diagnosis or intensified imaging follow-up with advanced sequences such as perfusion and diffusion MRI.

The study’s limitations are those inherent to its design. Thirty-six patients from a single institution, all of whom underwent surgery and therefore represent the more severe end of the non-response spectrum, is a modest and selected sample. The pathological endpoint was available only for lesions that failed treatment, so the volumetric signatures of well-behaved responding lesions could not be compared against a full spectrum of outcomes. The authors also note that edema trajectories are variable across patients, and the FLAIR-to-enhancing ratio trended, rather than decisively separated, the two pathological groups at all time points, with the significant six-week FLAIR volume difference being the clearest single discriminator. Validation in larger, multi-institutional cohorts, ideally with prospective edema volumetry and standardized imaging protocols, will be essential before these thresholds inform treatment decisions.

Even so, the study adds a deceptively simple idea to the post-radiosurgery toolkit: the brain around the tumor is talking, and clinicians should start listening to it quantitatively. For two decades, surveillance after radiosurgery has centered on the enhancing lesion, with edema assessed qualitatively, if at all. This analysis demonstrates that perilesional FLAIR volume follows measurable, statistically characterizable trajectories that correlate with the enhancing compartment early after treatment and diverge in diagnostically informative ways when treatment fails. As automated segmentation matures and is folded into routine radiology workflows, the humble edema measurement that radiologists have long eyeballed as background noise may become one of the most actionable numbers on the post-radiosurgery scan, helping to spare patients with radiation necrosis from unnecessary surgery while accelerating definitive intervention for those whose tumors have truly returned.

Subject of Research: Volumetric MRI analysis of edema and enhancement changes after stereotactic radiosurgery for non-responding brain metastases

Article Title: Volumetric analysis of radiographic changes after stereotactic radiosurgery for treatment of non-responding brain metastases

Article References: Rajkumar, S., Moy, H., Ji, X., Aboian, M., & Chiang, V. (2026). Volumetric analysis of radiographic changes after stereotactic radiosurgery for treatment of non-responding brain metastases. Journal of Neuro-Oncology, 180(1), Article 7. https://doi.org/10.1007/s11060-026-05824-9

Image Credits: AI Generated

DOI: 10.1007/s11060-026-05824-9

Keywords: stereotactic radiosurgery, brain metastases, radiation necrosis, FLAIR volumetry, perilesional edema, Gamma Knife, MRI surveillance, tumor recurrence, neuro-oncology, volumetric analysis, Journal of Neuro-Oncology, Volumetric