The research, led by Valeriia Abramova of the Computer Vision and Robotics Institute at the University of Girona, together with neurologists at Hospital Universitari Dr Josep Trueta, addresses a gap that has limited existing AI stroke tools. Most current detection systems treat the problem as a simple yes-or-no question: is a large vessel occlusion present, or not? But clinicians need more than that. The choice of endovascular strategy differs depending on which vessel segment is occluded. Blockages in the posterior circulation, such as the basilar artery, are more often treated with balloon angioplasty and permanent stents, while occlusions in the anterior circulation typically call for stent thrombectomy or catheter aspiration. Clinical guidelines for posterior circulation occlusions are still evolving, and uncertainty persists over optimal anesthesia and treatment approaches. Knowing the precise vessel involved therefore shapes intervention planning, prognosis estimation, and clinical interpretation in ways a binary alarm cannot.

Commercial software packages such as RapidLVO, Viz-LVO, e-CTA, and StrokeViewer-LVO have brought automated detection into hospitals, but they remain constrained by reduced sensitivity for distal occlusions, vulnerability to imaging artifacts that trigger false positives, and dependence on standardized acquisition protocols that can hamper generalization across institutions. Recent research efforts have pushed toward explicit localization of occlusions, including two-stage convolutional neural network pipelines on four-dimensional CT angiography and self-configuring object detection frameworks applied to maximum intensity projection images. Yet most of these approaches stop at drawing a bounding box around the affected region without linking that region to a specific, named vessel segment. The Girona team’s contribution is to do both tasks at once, within a single unified framework.

Technically, the researchers adapted the state-of-the-art nnDetection framework, a self-configuring three-dimensional object detection method built on the Retina U-Net architecture. This network fuses the RetinaNet one-stage detector with a U-Net encoder-decoder, extracting multi-level image features through convolutional layers with instance normalization and ReLU activation, while lateral and transposed convolutional connections build a feature pyramid. Two sub-networks operate on the pyramid levels: one for classification and one for bounding box regression, trained respectively with binary cross-entropy loss and a generalized intersection-over-union loss, with additional semantic segmentation supervision during training. The classification branch labels each detected occlusion as belonging to one of three clinically crucial segments: the basilar artery, the terminal internal carotid artery, or the M1 segment of the middle cerebral artery. Model training used a fixed patch size of 128 by 128 by 128 voxels, sixty epochs, and stochastic gradient descent with Nesterov momentum, implemented in PyTorch on an NVIDIA GeForce GTX 1080 Ti GPU.

The study’s most distinctive twist lies in how it constrains the search space anatomically. The Circle of Willis is the ring-shaped arterial network at the center of the brain that supplies blood to cerebral tissue, and it encompasses precisely the vessels most often affected by these occlusions. Because manual segmentation of this structure is laborious and ground truth vessel masks were unavailable for the datasets, the team trained a standard nnU-Net segmentation model on data from the TopCoW challenge, a competition dedicated to topology-aware Circle of Willis segmentation in CT and MR angiography spanning diverse anatomical variants. Their segmentation model achieved a Dice score of 0.944 plus or minus 0.026. Applied to the stroke scans, the automatic Circle of Willis segmentation was expanded by 50 voxels in all three dimensions, a margin chosen to guarantee that all occlusions in the dataset fell inside the resulting region of interest.

This anatomical prior enabled a head-to-head comparison of two strategies. In the global approach, the network trained and inferred on the full CT angiography volume. In the local approach, images were cropped to the Circle of Willis-derived region before training and inference. The development dataset comprised 179 CT angiography scans acquired on a Philips Ingenuity scanner at Hospital Dr. Josep Trueta, all containing occlusions annotated with three-dimensional bounding boxes by an expert neurologist. Of these, 143 scans were used for training with five-fold cross-validation, while 36 were held out for internal testing. Class distribution reflected the natural epidemiology of these strokes: the M1 segment dominated at 67 percent of cases, terminal internal carotid artery occlusions comprised 25 percent, and basilar artery occlusions were rare at 8 percent.

The results on the internal test set were strikingly strong for both variants. The global approach achieved a detection sensitivity of 0.92 at 0.08 false positives per image, rising to 0.97 at 0.20 false positives per image, while the local approach matched the same sensitivities at just 0.03 and 0.13 false positives per image respectively. A case-level paired bootstrap analysis at a fixed operating point of 0.1 false positives per image found a mean sensitivity difference of 0.000 with a 95 percent confidence interval spanning negative 0.081 to positive 0.081, confirming no systematic performance gap between the strategies. Each approach missed only a single occlusion. Localization accuracy was equally tight: the mean three-dimensional distance between predicted and ground truth bounding box centers was 1.76 millimeters for both strategies. Interestingly, the full-volume model generated more low-confidence false positives, whereas the region-restricted model produced fewer spurious detections overall.

Classification performance held up nearly as well. On the internal test set, overall accuracy reached 94 percent for the global approach and 91 percent for the local approach, with Cohen’s kappa statistics of 0.88 and 0.82, both indicating almost perfect agreement with expert ground truth. Every basilar artery occlusion was classified correctly by both approaches, a success the authors attribute to the basilar artery’s distinctive, isolated position at the base of the brain, which reduces ambiguity. The main classification shortfall appeared in the terminal internal carotid artery class, where accuracy fell from 89 percent globally to 78 percent locally. Counting detection and classification together, the occlusion was both found and correctly labeled in 33 of 36 cases for the global approach and 32 of 36 for the local one.

Generalization was tested on the independent CODEC-IV benchmark, consisting of 48 CT perfusion-derived CT angiography scans from different scanners and hospitals. Detection sensitivity dropped to 0.71, but much of that decline traced to a mismatch in annotation conventions: the benchmark’s ground truth boxes were uniformly small and pinpointed the occlusion site, while the Girona training boxes captured the entire clot extent. When the true positive threshold was relaxed to 2 percent intersection over union, sensitivity climbed to 0.90 for the global approach and 0.88 for the local one, with mean center-to-center distances of roughly 2.4 millimeters confirming that the models were accurately placing their predictions even when box dimensions diverged. On a supplementary inference-only evaluation of the IACTA-EST challenge dataset, which included 51 occlusion-negative cases, the global model achieved an area under the curve of 0.95 in separating positive from negative cases, against 0.84 for the local model, hinting that a well-chosen confidence threshold could suppress false alarms in clinical triage.

Perhaps the most consequential number is temporal. Restricting inference to the Circle of Willis region made the local approach approximately 3.3 times faster than the global one, cutting mean processing time per case from about 227 seconds to roughly 89 seconds. In acute stroke care, where treatment delays of minutes measurably worsen neurological outcomes, that speedup matters even before accounting for the reduced computational cost of running on modest hospital hardware. The authors acknowledge limitations, including a single-center training cohort from one scanner, a small number of basilar artery cases, and class imbalance mirroring natural disease prevalence. Nonetheless, by coupling a deep object detection framework to an automatically derived vascular landmark, the study demonstrates that anatomical knowledge can be baked directly into machine learning pipelines, delivering detection and vessel-level diagnosis in one pass and pointing toward AI assistants that fit realistically into the breakneck rhythm of a stroke unit.

Subject of Research: Deep learning detection and vessel-level classification of large vessel occlusions in brain CT angiography guided by Circle of Willis localization

Article Title: Circle of Willis-Guided Localization for Simultaneous Detection and Classification of Large Vessel Occlusions in Brain CTA

Article References: Abramova, V., Oliver, A., Lal-Trehan Estrada, U. M., Hamadache, R. E., Martínez Arias, P., Freixenet, J., Terceño, M., Silva, Y., & Lladó, X. (2026). Circle of Willis-Guided Localization for Simultaneous Detection and Classification of Large Vessel Occlusions in Brain CTA. Neuroinformatics, 24(4), Article 62. https://doi.org/10.1007/s12021-026-09817-x

Image Credits: AI Generated

DOI: 10.1007/s12021-026-09817-x

Keywords: large vessel occlusion, stroke, CT angiography, Circle of Willis, deep learning, nnDetection, neuroinformatics, medical imaging, endovascular treatment, brain imaging, artificial intelligence, stroke diagnosis

The technology’s lineage traces back to the first-generation Merci retriever, a corkscrew-like device approved in the mid-2000s that proved mechanical clot removal in the brain was feasible but achieved only modest recanalization rates. The decisive turning point came with self-expanding stent retrievers, notably Solitaire and Trevo, which are compressed inside a microcatheter, navigated through the vasculature to the clot, and then unsheathed so the nitinol mesh expands and integrates with the thrombus. Unlike a static stent left in place, these retrievers engage the clot mechanically across its full length, allowing the operator to pull the entire construct into a guide catheter. Randomized trials published in 2015, including MR CLEAN, ESCAPE, EXTEND-IA, SWIFT PRIME, and REVASCAT, together demonstrated overwhelming benefit of endovascular thrombectomy over medical therapy alone, reshaping international guidelines almost overnight and establishing first-generation stent retrievers as the anchor of modern stroke intervention.

The technical principle behind the second-generation devices is deceptively simple: radial force from the expanding mesh compresses the clot against the vessel wall while individual struts penetrate the thrombus, creating a mechanical interlock. In practice, the interaction is governed by a complex interplay of clot composition, device geometry, and vessel size. Ischemic stroke thrombi vary enormously, from soft, red-cell-rich emboli shed from the heart to hard, fibrin-rich clots loaded with platelets and von Willebrand factor that resist mechanical integration. Histological analyses cited in the review show that fibrin-dense outer shells and platelet-rich regions correlate with failed retrieval and poorer revascularization outcomes. This biological heterogeneity has pushed engineers toward devices with segmented designs, larger mesh cells, and specialized capture zones. Multi-zone platforms such as NeVa incorporate discrete drop zones with tightly spaced struts designed to trap organized clots, while radially adjustable retrievers such as Tigertriever allow the operator to expand the device progressively until it matches the vessel diameter, an advantage in both oversized proximal vessels and narrow distal branches.

Device development has also converged on integrated retrieval-and-protection concepts. EmboTrap-class retrievers feature distal capture baskets intended to intercept fragments that would otherwise migrate downstream and cause new infarcts, a complication known as distal embolization. registries such as ARISE II and the global EXCELLENT registry for the EMBOTRAP device have reported high first-pass reperfusion rates with these hybrid designs. First-pass effect, meaning complete reperfusion achieved in a single retrieval attempt, has emerged as a key performance metric because each additional pass increases procedural time, trauma to the endothelium, and the risk of hemorrhagic transformation. Recent generation devices such as Solitaire X have demonstrated significantly improved first-pass success compared with their predecessors, and tip-design studies confirm that the shape and stiffness of the retriever’s distal end materially influence whether fragments escape during withdrawal.

In parallel with hardware evolution, procedural technique has advanced into highly choreographed combinations. The most influential refinement is the pairing of stent retrievers with large-bore aspiration catheters positioned at the face of the clot, a strategy variously branded as Solumbra, SAVE, or ARTS. Aspiration continuously extracts clot fragments dislodged by the retriever, reducing the shower of emboli that would otherwise travel into healthy territory. Balloon guide catheters add a second layer of protection by temporarily arresting antegrade flow in the parent artery, creating a stagnant zone from which debris can be vacuumed rather than washed distally. The randomized PROTECT-MT trial from China showed that balloon guide catheters significantly improve excellent reperfusion rates, validating what in-vitro flow studies had long predicted. For refractory occlusions, operators increasingly deploy double stent retrievers simultaneously, doubling the mechanical interface with the clot, and recent bench studies plus the randomized TWIN2WIN trial support this bail-out strategy, although cumulative vessel wall injury remains a documented concern in animal models.

The clinical indications for thrombectomy have expanded dramatically alongside the devices themselves. Landmark trials including DAWN and DEFUSE 3 extended the treatment window from six hours to twenty-four hours in patients selected by advanced perfusion imaging, demonstrating that brain tissue can remain salvageable long after symptom onset when collateral circulation is robust. More recently, attention has turned to posterior circulation strokes caused by basilar artery occlusion, which are uniformly devastating without treatment; trials such as ATTENTION and BAOCHE provided the first randomized evidence supporting endovascular therapy in this territory. Equally consequential are the new studies in large infarct cores, including SELECT2, ANGEL-ASPECT, and RESCUE-Japan LIMIT, which overturned the long-held exclusion of patients with extensive established damage and showed net functional benefit from thrombectomy even in these high-risk presentations.

The most recent frontier involves medium and distal vessel occlusions, blocks in arteries one to three millimeters in diameter that were historically managed with medication because standard devices were too bulky. Purpose-built low-profile retrievers, including 3-millimeter variants of Solitaire X and Trevo and the adjustable Tigertriever 13, have enabled operators to reach these small vessels, and a wave of randomized trials in 2025 and 2026, including DISTALS, DISTAL, and DISCOUNT, has begun to establish benefit under imaging-guided selection. The review emphasizes that territory-specific engineering, from smaller delivery profiles to softer, more flexible distal architectures, is now the dominant axis of device innovation, with hybrid devices such as Aperio and specialized platforms for cerebral venous sinus thrombosis broadening the field further.

Materials science is contributing a quieter but potentially transformative layer of progress. Nitinol remains the workhorse alloy because its superelasticity allows dense crimping and atraumatic self-expansion, but its poor radiographic visibility complicates positioning, prompting coatings and design changes that enhance fluoroscopic contrast. Surface engineering aims to reduce thrombogenicity and endothelial damage, with heparin-based hydrogel coatings, endothelium-mimicking bioactive layers, and nanostructured oxide films under investigation. More provocative are clot-adhesive coatings that deliberately bind to fibrin, effectively welding the retriever to resistant thrombi, and micro-patterned surfaces that increase contact area. In a striking departure from conventional designs, milli-spinner thrombectomy, reported in Nature in 2025, uses a rotating, tangle-forming structure to compress and extract clots regardless of composition, hinting that the retrieval paradigm itself may not be permanent.

Looking forward, the review identifies thrombus characterization as the bridge between biology and device choice. Radiomic analysis of clot appearance on imaging, combined with biomarkers of clot composition, could soon allow operators to predict before the first pass whether a given occlusion will yield to a standard retriever or demand an adjustable device, dual-stent technique, or direct aspiration. Personalized device selection of this kind would attack the core unresolved problems: fibrin-rich resistant thrombi, embolic complications, vascular injury from repeated passes, and the limited high-level evidence supporting many of the newest devices, which have largely been validated in registries rather than randomized trials. The authors argue that ongoing integration of engineering innovation with rigorous clinical data will support increasingly individualized and safer thrombectomy strategies, and with stroke remaining a leading cause of disability globally, even incremental gains in first-pass success translate into meaningful reductions in death and dependence. The stent retriever, born from a simple wire mesh, continues to evolve into a precision instrument tailored to the specific clot, vessel, and patient standing between a stroke and recovery.

The stakes of these technical refinements are best understood against the sheer scale of the disease. Global burden analyses cited in the review estimate that stroke affected well over a hundred million people worldwide in recent years, and large vessel occlusions contribute disproportionately to death and dependence because the entire territory of a major cerebral artery is threatened within minutes of onset. Intravenous thrombolysis, the other pillar of acute reperfusion therapy, dissolves clot biochemically but achieves recanalization in only a minority of large vessel occlusions and carries a risk of arterial reocclusion, which is why mechanical retrieval became indispensable.

The review also situates current practice within the 2026 American Heart Association and American Stroke Association guideline for early management of acute ischemic stroke, reflecting how trial evidence is rapidly codified into standards of care. Beyond the procedure itself, the authors note that reperfusion initiates a second wave of injury, including blood-brain barrier breakdown and neuroinflammation, meaning that restoring flow is necessary but not always sufficient for good functional recovery. This biological reality underscores why procedural metrics such as first-pass success and reduced embolization matter clinically, and why the field increasingly views mechanical thrombectomy not as an isolated engineering problem but as one component of a broader effort spanning imaging selection, device design, and post-reperfusion neuroprotection.

Subject of Research: Technological advances and clinical applications of stent retrievers in endovascular thrombectomy for acute ischemic stroke.

Article Title: Stent retrievers for acute ischemic stroke: technological advances, clinical applications, and future perspectives

Article References: Xie, Z., Wang, Z., Fu, P., Shi, Z., Zhuang, Z., Wang, H., Xiang, Y., Yin, X., & Chen, Z. (2026). Stent retrievers for acute ischemic stroke: technological advances, clinical applications, and future perspectives. Journal of Neurology, 273(10), Article 590. https://doi.org/10.1007/s00415-026-14126-z

Image Credits: AI Generated

DOI: 10.1007/s00415-026-14126-z

Keywords: acute ischemic stroke, stent retriever, mechanical thrombectomy, large vessel occlusion, endovascular treatment, first-pass reperfusion, nitinol, distal embolization, balloon guide catheter, medium vessel occlusion, thrombus composition, reperfusion