Orthopedics · Trauma radiograph fracture detection
BoneView
Gleamer
The most widely studied fracture-detection AI in emergency radiology, and a lesson in reading its own evidence carefully. The field-defining reader studies — a 24-reader trial in Radiology that won the RSNA's Margulis award — were designed and funded by the manufacturer and show a real jump in sensitivity when clinicians read with the tool. The most recent independent systematic review is more measured: BoneView may raise sensitivity, but its effect on specificity and overall accuracy remains uncertain. What is not in doubt is reach, with clearances across the US, EU and Canada and a place in England's NICE early-value assessment of fracture-detection AI.
Performance Metrics
Clinical Evidence
BoneView has the largest peer-reviewed footprint of any fracture-detection AI, and the honest reading separates the manufacturer's own studies from the independent ones. The two field-defining papers were authored and funded by Gleamer. Guermazi and colleagues (Radiology 2022) ran a multi-reader multi-case study of 480 examinations with 24 readers spanning six clinician types — radiologists, orthopaedic surgeons, emergency physicians, physician assistants, rheumatologists and family physicians — and reported a 10.4 percentage-point gain in patient-wise sensitivity when reading with support, a specificity gain of about five points and a mean reading time shortened by roughly six seconds per examination; the study won the 2022 RSNA Alexander R. Margulis Award, and five of its authors are Gleamer employees. The earlier Duron study (Radiology 2021), across 600 adults, 17 French centres and 12 readers, reported a sensitivity gain of 8.7 points and a specificity gain of 4.1 points. Both are real results, and both come from the device's maker. The independent evidence is more restrained and is the fairer basis for a procurement decision. Bousson and colleagues (Academic Radiology 2023) ran a real-world emergency-department head-to-head of three commercial algorithms — BoneView among them — across 1,500 radiographs from 1,210 patients, judged against a panel of four musculoskeletal radiologists; the value is the neutral, competitive framing rather than a single headline number. Kwee and Kwee's systematic review (European Journal of Radiology 2025) pooled eight studies and concluded that BoneView may improve sensitivity but that its effect on specificity and overall accuracy remains uncertain, flagging a high risk of bias in patient selection in five of the eight — the single most important sentence in the whole evidence base to carry forward. A prospective crossover reader study from a Munich group (Prucker et al., International Journal of Medical Informatics 2026) found the clearest reproducible benefit was not accuracy but speed: shorter interpretation times, without a significant change in fracture, dislocation or effusion accuracy. Paediatric performance rests on a smaller external validation (Nguyen et al., Skeletal Radiology 2022) of 300 examinations in children aged two to twenty-one.
| Study | Design | n | Sensitivity | Specificity | AUC | Published |
|---|---|---|---|---|---|---|
| Guermazi A, Tannoury C, Kompel AJ, et al. (five Gleamer co-authors; manufacturer-funded — 2022 RSNA Margulis Award) | RetrospectiveRetrospective | 480 | +10.4 pts with AI (patient-wise) | +5.0 pts with AI | — | Radiology, 2022 (PMID 34931859); multi-reader multi-case study, 24 readers across six clinician types; mean reading time also reduced ~6.3 s/exam |
| Duron L, Ducarouge A, Gillibert A, et al. (Ducarouge is a Gleamer co-founder — manufacturer-authored) | RetrospectiveRetrospective | 600 | +8.7 pts with AI (P=.006) | +4.1 pts with AI (P=.03) | — | Radiology, 2021 (PMID 33944629); 17 French centres, 12 readers (6 radiologists + 6 emergency physicians) |
| Regnard N-E, et al. (first author is a Gleamer co-founder — manufacturer-affiliated) | RetrospectiveRetrospective | 4,774 | 98.1% lesion-wise (fractures); dislocations 89.9%, elbow effusions 91.5%, focal bone lesions 98.1% | — | — | European Journal of Radiology, 2022 (PMID 35921795); 14 centres — this study defines the four-finding CE scope, broader than the US label |
| Bousson V, et al. (independent — real-world ED head-to-head of three commercial algorithms) | RetrospectiveRetrospective | 1,210 | — | — | — | Academic Radiology, 2023 (PMID 37468377); 1,500 radiographs from 1,210 patients; BoneView vs SmartUrgence vs Rayvolve, ground truth by four MSK radiologists across 13 body regions |
| Nguyen T, Ducarouge A, et al. (external validation, paediatric — Gleamer co-authors) | RetrospectiveRetrospective | 300 | e.g. forearm 92.9% | e.g. forearm 98.1% | — | Skeletal Radiology, 2022 (PMID 35522332); acute appendicular fractures, children aged 2–21, 150 fracture / 150 non-fracture, 60 per body part |
| Prucker P, Busch F, et al. (independent — TUM Munich) | ProspectiveProspective | 0 | — | — | — | International Journal of Medical Informatics, 2026 (PMID 40975012); prospective crossover reader study, three commercial algorithms; reduced interpretation time (e.g. 34s→21–25s), no significant change in fracture/dislocation/effusion accuracy |
| Kwee RM, Kwee TC (independent — systematic review) | RetrospectiveRetrospective | 0 | — | — | — | European Journal of Radiology, 2025 (PMID 40554942); 8 studies — BoneView may improve sensitivity, effect on specificity and overall accuracy uncertain; high risk of patient-selection bias in 5 of 8 |
Clinical Pulse
Verified clinician sentiment
Verified clinician reviews launching soon · Apply to contribute →
Reviews are signed by clinicians verified against their regulator's public register.
Inside the algorithm
Editorial featureHow BoneView reaches its output.
Five stages — from raw input to verdict — drawn from manufacturer documentation and the public regulatory record.
- INGEST
- NORMALISE
- DETECT
- LOCALISE
- VERDICT
Stage 01 · INGEST
Trauma radiographs arrive from the imaging workflow.
BoneView receives musculoskeletal radiographs as studies routed from the modality or , most often in an emergency or trauma setting. Analysis runs automatically in the background as studies arrive, in parallel to the clinician's read; the source images are not altered and no additional imaging is required. Distribution through imaging-platform partners means the tool typically appears inside an existing PACS or worklist rather than as a separate application.
Input
Radiograph DICOM
Inference
Background, per-study
Inside the Auris+ Listing
Five more sections complete this device’s Auris+ Listing.
Decision Ledger
ProPro unlocks a private, cross-vendor log of every case you read on this device — what the AI called, what you concluded, and a one-line reason if you overrode it. Ready for the EU AI Act's deployer logging obligations when they land in 2028.
Clinical Evidence Deep Dive
ProPro unlocks the structured clinical-evidence summary — study count, target patient population, and a tabular accuracy-metrics view drawn from peer-reviewed sources.
Peer-Reviewed Publications
ProPro unlocks the curated peer-reviewed publication list with PubMed cross-links — the citation backbone of every editorial verdict.
Post-Market & Regulatory Conditions
ProPro unlocks the post-market surveillance summary, recall record, and the conditions of approval that bound real-world use.
AI Algorithm Version History
ProPro unlocks the chronological record of algorithm version changes — what changed when, drawn from manufacturer changelogs and regulatory filings.
Regulatory Approvals
K222176
Class II
Safety Record
No recalls, field safety notices or FDA safety communications attributable to Gleamer or BoneView were found in indexed public sources as of July 2026 — read as "none found", not an exhaustive audit, since the FDA recall and MAUDE databases could not be queried directly from this environment. The clinically meaningful caveat is not a recall but the gap between manufacturer-reported and independently reproduced performance: a fracture-detection aid that does not mark a region is not a normal read, and the independent systematic review is explicit that the specificity and accuracy effects are uncertain. The US and EU labels also differ in scope — the additional dislocation, effusion and bone-lesion findings are CE-marked, not FDA-cleared — so what the tool is permitted to report depends on the jurisdiction it is deployed in.
Intended Use & Indications
BoneView is a software-only device that analyses digital radiographs for signs of acute fracture and presents its findings as region marks alongside the clinician's own interpretation. The scope differs by jurisdiction, and the difference matters for procurement. Per the FDA 510(k) record the US indication is fracture detection: in adults it covers the limbs, pelvis, rib cage and dorsolumbar (thoracolumbar) spine; a second clearance in 2023 extended the indication to paediatric patients aged two and older, but only on limb radiographs. The broader four-finding capability marketed in Europe — fractures plus dislocations, elbow joint effusions and focal bone lesions — is the CE-marked scope, evaluated in the manufacturer's wider validation work, and should not be read into the narrower US clearance. Across every configuration BoneView is an adjunctive, concurrent-reading aid: the treating clinician remains responsible for the diagnosis.