The clinical portion of the study focused on patients evaluated for a first episode of lower-extremity deep vein thrombosis between May and October 2022. Each patient underwent complete whole-leg ultrasonography and a formal pulmonary embolism assessment, using either computed tomography pulmonary angiography or pulmonary perfusion imaging. The clots were sorted into four mutually exclusive categories: isolated muscular calf vein thrombosis confined to the gastrocnemius or soleal veins; isolated axial calf thrombosis confined to the anterior tibial, posterior tibial, or peroneal veins; isolated proximal thrombosis from the popliteal vein upward; and combined muscular calf and proximal involvement. Because several categories contained few patients, the team compared pulmonary embolism rates across all four groups with a Fisher-Freeman-Halton exact test rather than conventional chi-square analysis.

The headline numbers are striking. Among the 160 patients, 30 had concurrent pulmonary embolism, an overall prevalence of 18.8 percent. For the 125 patients whose clots were confined to the muscular calf veins, the rate was 11.2 percent, with an exact 95 percent confidence interval of 6.3 to 18.1 percent. Isolated axial calf thrombosis produced pulmonary embolism in 1 of 4 patients, isolated proximal thrombosis in 3 of 10, and combined muscular calf and proximal thrombosis in 12 of 21, or 57.1 percent. The overall distribution differed across groups with an exact p value of 0.000020. The authors are careful to note that the small axial and proximal subgroups carry very wide confidence intervals, so no individual pairwise comparison between categories can be declared significant. Still, the pattern suggests that the more anatomical territory a clot occupies, and particularly the combination of muscular calf and proximal involvement, the more likely the lungs are already involved.

That clinical finding alone would be noteworthy, but the study’s second half is what gives it an unusual edge. Rather than treating the calf as an opaque black box, the team built a three-dimensional structural finite-element model from a single de-identified computed tomography angiography dataset taken from an individual without thrombosis. The scan covered the abdomen and lower limbs at one-millimeter slice thickness. The researchers segmented the data in Materialise Magics, smoothed the surfaces in Geomagic, and imported the geometry into COMSOL Multiphysics version 6.0 for a stationary structural analysis. The modeled region extended from knee to ankle and included the principal bones, the gastrocnemius and soleus muscles and their tendons, vascular structures, and surrounding soft tissue. The gastrocnemius and soleal veins were represented as solid, low-stiffness structures with a nominal diameter of three millimeters, and muscle fibers were oriented at 16.2 degrees relative to the tibial axis.

The physics of the model hinged on a simple observation from elastography experiments: muscle stiffens dramatically when it contracts. Relaxed muscle was assigned an elastic modulus of 8.2 kilopascals with a Poisson ratio of 0.49, while contracting muscle was given a modulus of 200 kilopascals, roughly a twenty-five-fold increase, with a Poisson ratio near 0.50. Contraction was implemented as a 20 percent external material strain in whichever muscle domain was active. Three loading conditions were tested: gastrocnemius contraction with the soleus relaxed, soleus contraction with the gastrocnemius relaxed, and simultaneous co-contraction of both muscles. The geometry was partitioned into more than 1.3 million tetrahedral elements spanning nearly 224,000 mesh vertices, with roller supports on the inner soleus surface and bone-related constraints anchoring the structure. Muscle and adjacent venous structures were treated as ideally connected, with no relative sliding at their interface.

The simulation delivered both a global answer and a local map. At the final recorded loading point, mean muscle von Mises stress reached 21.51 kilopascals during gastrocnemius contraction, only 1.16 kilopascals during soleus contraction, and 9.68 kilopascals during co-contraction. Total modeled venous-domain volume, a proxy for venous compression, shrank by 49.25 cubic millimeters under gastrocnemius loading, 17.07 under soleus loading, and 64.08 under co-contraction. In other words, the biggest squeeze on the deep veins came not from either muscle acting alone but from the two firing together, which is close to what happens during real movements such as rising onto the toes. The deformation was decidedly nonuniform: stress and strain maps concentrated the strongest structural response around proximal venous branching regions and the adjacent intramuscular venous segments, rather than distributing it evenly across the venous tree.

That spatial localization matters because it converges with decades of anatomical and autopsy work. The soleal venous network is known to be a dense, branching web of intramuscular channels with defined proximal drainage routes, and prior autopsy studies of fatal pulmonary embolism have repeatedly identified the soleal veins and their drainage pathways as important origins of embolizing thrombus. The new model cannot prove that contraction detaches clots, since it deliberately excluded blood flow, intraluminal pressure, venous valves, fluid-structure interaction, and any thrombus material. What it does establish is the mechanical substrate: contracting calf muscle demonstrably transfers load to the deep veins, the magnitude of compression depends on which muscles are active, and the regions of greatest deformation coincide exactly with the branching zones where thrombi are known to form and from which they are known to travel.

The authors pair the simulation with illustrative ultrasound images showing the soleal network as multiple communicating intramuscular channels with segmentally dilated venous spaces, alongside dynamic color Doppler acquisitions that capture flow signals during calf-pump maneuvers. Together, the imaging and the mechanics support a concrete clinical proposal: standardized dynamic venous ultrasonography, in which venous caliber and flow are assessed at rest and during controlled calf-muscle activation. The high-response regions identified by the model, especially proximal branching areas and adjacent intramuscular segments, provide anatomically defined targets for such scanning. This is particularly relevant because muscular calf vein thrombosis can be asymptomatic, can coexist with silent pulmonary embolism, and dominated the study cohort at 78.1 percent of all cases.

The limitations are substantial and the authors state them plainly. The clinical analysis was retrospective, single-center, and built on four markedly unequal groups, with only 30 pulmonary embolism events in total and no patient-level data on coagulation markers, thrombus dimensions, occlusion grade, treatment, or follow-up, precluding any adjustment for confounding. The temporal sequence of leg and lung imaging does not establish where an embolus originated. The finite-element component used one normal-geometry model, idealized linear materials, and a nominal venous diameter, with no mesh-independence study, sensitivity analysis, or experimental validation; the archived record did not even retain a separate numerical venous modulus. A thrombosed-vein model, requiring disease-specific geometry and material properties, was explicitly out of scope. The simulation therefore serves as a normal-anatomy structural reference, not a hemodynamic or embolic prediction.

Even within those constraints, the study marks a genuine step toward treating the calf’s muscular veins as biomechanically active structures rather than anatomical afterthoughts. It quantifies, in numbers and in maps, the squeeze that walking places on the very vessels where a large share of dangerous clots begin, and it ties that squeeze to a measurable gradient of pulmonary embolism risk across clot locations. The authors call for prospective studies recording coagulation markers, thrombus dimensions, treatment exposure, and serial imaging before and after treatment, with the present model serving as a biomechanical foundation rather than a treatment rule. Until such validation arrives, the practical message is modest but clear: when clinicians scan the legs, the small muscular veins of the calf deserve the same careful, dynamic attention as the large axial trunks, because the mechanical forces acting on them with every step are real, localized, and now finally visible.

Subject of Research: Anatomical distribution of lower-extremity deep vein thrombosis and calf muscle-vein compression mechanics analyzed with finite-element modeling

Article Title: Anatomical distribution of lower-extremity deep vein thrombosis and pulmonary embolism: a retrospective study with finite-element analysis of calf muscle-vein mechanics

Article References: Hong, X., Hou, J., & Hong, X. (2026). Anatomical distribution of lower-extremity deep vein thrombosis and pulmonary embolism: a retrospective study with finite-element analysis of calf muscle-vein mechanics. Frontiers in Cardiovascular Medicine, 13, Article 1730081. https://doi.org/10.3389/fcvm.2026.1730081

Image Credits: AI Generated

DOI: 10.3389/fcvm.2026.1730081

Keywords: deep vein thrombosis, pulmonary embolism, muscular calf veins, gastrocnemius, soleus, finite-element analysis, venous compression, whole-leg ultrasonography, venous thromboembolism, biomechanics, CT angiography, retrospective study