The clinical motivation is stark. Osteoporotic vertebral compression fractures are among the most common skeletal injuries in older adults, yet they frequently escape detection because patients are rarely evaluated for bone health when they come in for unrelated imaging. Many fractures are asymptomatic or produce only vague symptoms, and dedicated osteoporosis screening remains underutilized due to cost, access barriers, and the logistical burden of separate appointments. Opportunistic screening — extracting diagnostic value from imaging already acquired for other purposes — has emerged as a pragmatic way to close this care gap, and the Beirut team’s system is among the first to combine fracture detection, severity grading, and Schmorl’s node identification in a single deployed workflow.

Technically, the pipeline operates in two stages, both built on the YOLOv11 architecture, a recent generation of the You Only Look Once family of real-time computer vision models. The first stage is a segmentation model, based on the YOLOv11l-seg variant, that was trained on three public datasets — CTSpine1K, VerSe2019, and VerSe2020 — totaling nearly 1,500 CT volumes with expert-labeled vertebrae. After harmonization through isotropic resampling and standardized anatomical labeling, the model learned to localize, segment, and individually label every vertebra from C1 down to S1. Unlike U-Net-style semantic segmentation networks, YOLOv11 natively handles detection, segmentation, and classification in a single pass, which the researchers say was critical for the speed demanded by live clinical deployment.

The segmentation results are striking. When at least one end of the spine was visible in the scan, the model achieved 96 percent Top-1 accuracy in assigning the correct anatomical label to each vertebra, and even when both spinal extremities were absent from the field of view, accuracy only dropped to 86 percent. Top-3 accuracy — meaning the correct label appeared among the model’s three highest-ranked predictions — reached a perfect 100 percent in both conditions. The mean Intersection over Union, a standard measure of how closely predicted segmentation masks match ground truth, averaged 0.94, indicating near-perfect spatial overlap.

Once each vertebra is isolated and cropped, it flows into a multi-stage classification pipeline trained on a retrospectively curated institutional dataset of 5,000 CT scans from the American University of Beirut Medical Center, collected between 2016 and 2020 from patients aged 65 and older who underwent abdominal, pelvic, or lumbar imaging for non-spinal reasons. Two experienced radiologists annotated every vertebra using a custom-built DICOM labeling tool, applying the Genant semi-quantitative grading method — the clinical gold standard that classifies fractures as mild, moderate, or severe based on the degree of vertebral height loss. An initial consensus phase on 500 scans calibrated the annotators before the remaining scans were divided between them.

The classification cascade itself is elegantly layered. A primary classifier first assigns each vertebra to one of three categories: normal, fractured, or unknown — the latter serving as a safety valve for cases obscured by motion artifacts, surgical hardware, or severe degenerative change, which are automatically routed to a radiologist for mandatory review. Vertebrae flagged as fractured then pass to a severity grader, while all confidently classified vertebrae are additionally screened for Schmorl’s nodes by a separate binary detector. The outputs are aggregated per vertebra as multi-label results, allowing clinicians to see, for example, both a moderate fracture and a co-occurring Schmorl’s node on the same vertebral body.

Performance on held-out test data was strong across the board. The primary fracture classifier reached 92.7 percent vertebra-level accuracy, with 92.6 percent sensitivity, 95.8 percent specificity, and an F1-score of 91.6 percent for fractured vertebrae — figures that match or exceed most published approaches, including a commercially deployed algorithm from Zebra Medical that achieved only 68 percent sensitivity in a real-world evaluation. Schmorl’s node detection achieved 91.9 percent accuracy with an F1-score of 93.5 percent for the Schmorl class, which the authors believe is the first reported benchmark for automated Schmorl’s node detection from CT. Severity grading proved harder, at 85.0 percent overall accuracy, with the moderate category the weakest link — a limitation the team attributes largely to the opportunistic nature of the scans, which were not acquired with bone-optimized protocols and therefore carry reduced vertebral resolution.

End-to-end evaluation of the full pipeline, from raw CT input through segmentation, classification, and web-based visualization, showed an overall success rate of 85 percent — defined as cases where every vertebra was correctly segmented and classified — and 90 percent multi-label consistency with radiologist consensus. To make the system interpretable rather than a black box, the researchers applied Eigen-CAM, a gradient-free technique that computes principal components of the final convolutional feature maps to produce spatial heatmaps. These visualizations revealed clinically meaningful patterns: normal vertebrae produced compact central activations, mild fractures shifted attention toward the superior endplate, moderate fractures spread diffusely across the vertebral body, severe fractures concentrated sharply at the site of collapse, and Schmorl’s nodes appeared as focused circular hotspots at the endplate — mirroring the very progression described by the Genant grading system.

Perhaps the most distinctive feature is what happens after deployment. The system is live at AUBMC, integrated into routine radiology workflows through a React-based web platform where radiologists can review every per-vertebra prediction, view confidence scores, and override results with a single click. Every correction is logged and stored alongside the original model output, creating a continuously growing pool of expert-validated annotations. These feed scheduled retraining cycles — with strict separation between training and evaluation data to prevent leakage, and systematic benchmarking before and after each retraining to verify genuine improvement — meaning the model steadily adapts to the institution’s real data distribution over time.

The team deliberately trained and tested on messy, heterogeneous, real-world imaging rather than pristine curated datasets, incorporating variability in scanner vendors, reconstruction kernels, slice thickness, contrast administration, and patient positioning. That choice makes the reported numbers a more honest reflection of clinical performance, and it positions the pipeline as a practical blueprint for bringing AI-assisted vertebral screening into everyday radiology. Looking ahead, the researchers plan to extend the system to MRI and DEXA imaging, fuse clinical data for personalized fracture risk assessment, and use large language models to automate structured reporting — steps that could transform a routine chest scan into a comprehensive, opportunistic bone health checkup for millions of older adults.

Subject of Research: Deep learning detection of vertebral fractures and Schmorl's nodes from routine CT scans

Article Title: End-to-end deep learning pipeline for opportunistic vertebral fracture and Schmorl’s node detection from routine CT scans

Article References: Hasan, H., El-Hajj Fuleihan, G., Natout, M., Mohanna, A., Tajeddine, R., Rafeh, W., Saleh, F., El Alam, R., Chahine, R., Khoury, N., & Chehab, A. (2026). End-to-end deep learning pipeline for opportunistic vertebral fracture and Schmorl’s node detection from routine CT scans. Frontiers in Musculoskeletal Disorders, 4, Article 1718482. https://doi.org/10.3389/fmscd.2026.1718482

Image Credits: AI Generated

DOI: 10.3389/fmscd.2026.1718482

Keywords: deep learning, vertebral fracture, opportunistic screening, osteoporosis, CT imaging, YOLOv11, Schmorl's nodes, medical AI, End-to-end, deep, learning, pipeline

The research team, led by Bing-Hao Lu, Wei-Han Lin, and Kuo-Yuan Huang, focused on a group of patients who are, in many respects, the clinical frontline of the osteoporosis epidemic: 230 hospitalized adults who had undergone surgery for fragility fractures of the hip or spine. These are the fractures that occur when a fall from standing height, or even a trivial mechanical insult, is enough to break bone weakened by age-related mineral loss. Hip fractures in particular carry a grim prognosis, with prior meta-analyses showing excess mortality that persists for years after the injury, alongside substantial long-term disability. Identifying better tools to stratify skeletal risk in such patients is therefore not an academic exercise but a pressing clinical need, because the first fragility fracture is one of the strongest predictors that a second one will follow.

Abdominal aortic calcification offers an attractive candidate marker because it can be observed on imaging that many of these patients already undergo. The researchers quantified AAC on lateral spine radiographs using the 24-point Kauppila score, a well-established semiquantitative index that grades the severity of calcified deposits along both the anterior and posterior walls of the aorta across four lumbar vertebral segments. By summing the scores at each level, the method yields a continuous measure of calcification burden ranging from zero to 24. The technique has a long pedigree: it was first developed in the Framingham Heart Study population, where aortic calcific deposits were shown to predict vascular morbidity and mortality, and it has since been adapted for use with dual-energy X-ray absorptiometry images, potentially allowing osteoporosis screening visits to double as vascular assessments.

The biological rationale for a bone-vascular link is compelling. Both bone mineralization and vascular calcification involve the deposition of calcium phosphate crystals, but they sit at opposite ends of a regulatory spectrum. Cells in the vascular wall, under conditions of oxidative stress, inflammation, disordered mineral metabolism, and aging, can adopt an osteoblast-like phenotype and actively lay down bone-like matrix within the artery. Meanwhile, the same molecular pathways, including osteoprotegerin, bone morphogenetic proteins, and vitamin K–dependent proteins such as matrix Gla protein, participate in regulating both processes. Epidemiological evidence has repeatedly connected the two: lower bone mineral density has been associated with coronary artery calcium, and aortic calcification has been linked to bone loss over 25-year follow-up in the Framingham cohort. Systematic reviews and meta-analyses of observational studies have reported that higher AAC burden is associated with lower bone mineral density and increased fracture risk in general populations.

Against this backdrop, the Taiwanese team set out to test whether AAC could meaningfully stratify skeletal outcomes in patients who had already fractured. They examined associations between AAC burden and bone mineral density measured by dual-energy X-ray absorptiometry, with particular attention to the femoral neck, the site most relevant to hip fracture. They also explored relationships with FRAX-estimated fracture probabilities, the widely used algorithm that integrates clinical risk factors with or without bone density to estimate ten-year fracture likelihood. Fracture severity was graded using established classification systems, including the semiquantitative method for vertebral fractures and the Garden and Evans classifications for femoral neck and trochanteric hip fractures, respectively. Finally, the investigators tracked refracture over time using Kaplan-Meier survival curves and Cox proportional hazards models, anchoring all follow-up to a common baseline defined as the latest of three dates: the index operation, the bone density scan, and the radiographic AAC assessment.

The headline result was a genuine but fragile association. In unadjusted analyses, higher AAC scores were significantly correlated with lower femoral neck bone mineral density, with a Spearman correlation coefficient of −0.230 and a p-value below 0.001. This is consistent with the broader literature and suggests that, at the crude level, calcified arteries and demineralized femoral necks do travel together in fracture patients, just as they do in community cohorts. However, when the researchers adjusted for potential confounders in multivariable regression models, the association was attenuated and lost its independent statistical footing. In other words, much of the apparent relationship between aortic calcification and low bone density in these patients could be explained by factors that AAC and bone density share in common: advancing age, sex, and body-composition characteristics.

The same pattern of attenuated significance held for the other skeletal outcomes. Although AAC showed weak positive correlations with FRAX-estimated probabilities of major osteoporotic and hip fracture, it did not independently predict the severity of the vertebral or hip fractures that brought these patients to the operating room in the first place. A more severe calcification burden did not translate into a more comminuted fracture, a higher-grade compression deformity, or a worse anatomic pattern. This finding cuts against the intuitive expectation that vascular disease and skeletal fragility would progress in lockstep, even within a population already selected for having sustained a fragility fracture.

Perhaps the most clinically consequential result concerned refracture risk. Among the 224 patients who could be followed from the common baseline, 24 experienced a first refracture during the observation period. Refracture-free survival, plotted by Kaplan-Meier analysis, did not differ between patients with higher and lower AAC burdens, and the log-rank test yielded a p-value of 0.982, indicating near-identical curves. In the multivariable Cox model, high AAC was not associated with refracture risk, with a hazard ratio of 0.90 and a 95 percent confidence interval of 0.38 to 2.15, spanning unity comfortably. The authors are appropriately cautious here, noting that the refracture analysis should be interpreted as exploratory because the number of events was limited, which widens confidence intervals and reduces the study’s power to detect modest effects. Still, the direction of the result offers no encouragement that AAC adds predictive value in this setting.

The broader significance of the study lies in what it says about context dependence in biomarker research. Most of the evidence linking AAC to skeletal outcomes comes from community-dwelling populations undergoing routine osteoporosis screening, where recent work, including automated AAC quantification combined with trabecular bone score, has suggested independent fracture prediction. But patients who have already fractured and require hospitalization and surgical fixation represent a different clinical universe: they are older, frailer, and closer to the severe end of the bone-vascular disease spectrum. Within such a selected cohort, the variance in skeletal health attributable to vascular calcification may already be absorbed by the very fact of the fracture, and by the demographic and body-composition factors that drove both processes. The authors conclude that AAC showed limited independent value for skeletal stratification in this inpatient cohort beyond shared demographic and body-composition factors. For clinicians, the message is that a calcified aorta glimpsed on a lateral spine film should not be treated as a surrogate for a fragile skeleton or a forthcoming refracture in fracture patients, even as it retains well-documented value as a marker of cardiovascular risk. For researchers, the study is a reminder that biomarkers validated in screening populations must earn their place, study by study, in the more complex terrain of acute fracture care.

One practical implication concerns how AAC is measured in routine care. Because the Kauppila scoring system can be applied to lateral spine images acquired during standard densitometry, vascular calcification assessment requires no additional radiation exposure or cost, which explains much of the enthusiasm for integrating it into fracture risk workflows. The present findings temper that enthusiasm for inpatient populations, but they do not negate the score’s established role in cardiovascular prognostication, where meta-analytic evidence links aortic calcification to incident vascular events and mortality.

Several design features of the study merit consideration when weighing its conclusions. The retrospective single-center design means that AAC burden was quantified from radiographs obtained for clinical rather than research purposes, and the timing of imaging relative to the index fracture may have varied across patients. The modest sample size of 230, while respectable for an inpatient fracture cohort, limits the precision of adjusted estimates, and the 24 refracture events provide only coarse resolution for survival modeling. Selection factors inherent to hospitalization, including the decision to obtain lateral spine imaging and DXA, may also have shaped the cohort in ways that are difficult to quantify. Future prospective studies with standardized imaging protocols, larger event counts, and longer follow-up will be needed to determine whether AAC retains any incremental prognostic value once fracture has occurred, or whether its utility remains confined to community screening settings.

Subject of Research: The association between abdominal aortic calcification and bone mineral density, fracture severity, and refracture risk in hospitalized patients with surgically treated fragility fractures.

Article Title: Abdominal aortic calcification and skeletal outcomes in hospitalized patients with surgically treated fragility fractures: a retrospective cohort study

Article References: Lu, B.-H., Lin, W.-H., & Huang, K.-Y. (2026). Abdominal aortic calcification and skeletal outcomes in hospitalized patients with surgically treated fragility fractures: a retrospective cohort study. Archives of Osteoporosis, 21(1), Article 137. https://doi.org/10.1007/s11657-026-01755-0

Image Credits: AI Generated

DOI: 10.1007/s11657-026-01755-0

Keywords: abdominal aortic calcification, bone mineral density, fragility fracture, refracture, osteoporosis, FRAX, Kauppila score, hip fracture, vertebral fracture, vascular calcification, dual-energy X-ray absorptiometry, retrospective cohort study