AI-powered opportunistic QCT
Opportunistic CT Bone Mineral Density with AI (mg/cm³)
Bone mineral density from routine CT — no extra scan, calibrated in mg/cm³. BMDNow transforms dual-energy CT (DECT) physics and machine learning into an automated osteoporosis screening layer—opportunistic, contrast-robust, multi-site (spine & hip), deployable across PACS and the cloud. Learn about Opportunistic CT Bone Mineral Density.
Opportunistic screening
Extract volumetric BMD from existing CT exams with high sensitivity/specificity for osteoporosis detection.
DECT material decomposition
Voxel-wise three-material separation (bone/water/fat) achieving calibrated mg/cm³, robust to IV contrast.
Clinician-ready outputs
Site-specific BMD, guideline categories, fracture risk, and longitudinal tracking with DICOM SR integration.
Why osteoporosis screening matters
Osteoporosis is underdiagnosed and undertreated despite its impact on fragility fractures and quality of life. Many eligible patients never undergo DXA; vertebral fractures carry excess mortality. Hospitals already produce large volumes of CT for other indications—leveraging those scans enables earlier detection without new appointments.
The problem
- Missed opportunities: many at-risk patients never receive DXA screening.
- Workflow friction: scheduling and capacity limit proactive screening.
- Fracture burden: vertebral and hip fractures raise morbidity and mortality.
The solution
- Opportunistic QCT extracts volumetric BMD from routine CT without extra scan time or dose.
- AI + DECT deliver calibrated mg/cm³ robust to contrast and marrow fat variability.
- Clinician-ready outputs (DICOM SR + PDF) fit existing reporting workflows.
Limitations of DXA (and why QCT helps)
- Projectional areal density can be confounded by degenerative change or body composition.
- Access/adherence barriers reduce real-world coverage.
- Volumetric, trabecular-focused QCT improves risk stratification.
How BMDNow computes opportunistic CT bone mineral density (QCT)
A standardized pipeline turns CT pixels into calibrated, clinically meaningful mg/cm³ values.
1) Ingest & QC
- DICOM import via PACS or secure web upload (TLS 1.3)
- Auto-detect region, kV pairs, and contrast phase using classifiers
- Quality flags for motion, coverage, and metal detection
2) Segment & Analyze
- AI segmentation of spine (L1–L4) and proximal femur
- DECT three-material decomposition (bone/water/fat)
- Single-kV mapping anchored to DECT truth when DECT unavailable
3) Compute BMD
- Voxel-wise volumetric BMD in mg/cm³ (trabecular-focused)
- Contrast-robust estimates excluding vascular voxels
- Confidence intervals and QC annotations (ACR QCT guidance)
4) Report & Integrate
- DICOM SR + PDF; optional HL7/FHIR
- Guideline categories: normal / osteopenia / osteoporosis
- Longitudinal comparisons for therapy monitoring
The science behind the numbers
Opportunistic QCT shows superior fracture prediction compared with DXA (AUC ~0.84 vs ~0.67) and higher case detection in multi-center cohorts. DECT provides energy-dependent attenuation for material decomposition; phantom-anchored linearity and repeatability have been demonstrated.
DECT physics
Two synchronized spectra enable material decomposition separating bone matrix, water, and fat at the voxel level.
- Vendor-calibrated coefficients per kV pair
- Phantom-anchored linearity and repeatability
- Trabecular focus aligned with fracture risk
AI & generalization
Segmentation and HU→BMD mapping trained across multi-vendor datasets; single-kV scans leverage DECT-anchored models to maintain calibrated estimates.
- Uncertainty estimates reported for clinical confidence
- Site-specific calibration options (phantom or asynchronous)
- On-prem or cloud deployment with audit trails
Clinical Evidence (overview)
- Fracture prediction: Opportunistic QCT AUC 0.84–0.94 vs ~0.67 for DXA.
- Case detection: Opportunistic QCT detects substantially more osteoporosis than DXA in population cohorts.
- Contrast robustness: Minimal difference between contrast and non-contrast in validated pipelines.
- Diagnostic thresholds: L1 attenuation norms in >20k adults to stage risk and guide action.
For radiology teams & health systems
Operational impact
- No extra scan time: opportunistic analysis from routine CT
- Automated reporting reduces radiologist effort per case
- Addresses under-screening where many eligible patients never receive DXA
Regulatory & privacy
- SaMD lifecycle aligned to MDR/FDA pathways
- GDPR/HIPAA-conscious processing with encryption & audit logs
- Role-based access and on-prem options
FAQ — BMDNow
Does BMDNow require dual-energy CT (DECT)?
BMDNow supports DECT for physics-first material decomposition. If only single-kV CT is available, BMDNow provides calibrated estimates via HU→BMD mapping anchored to a DECT ground-truth database.
Is intravenous contrast a problem?
The pipeline identifies and excludes vascular/contrast voxels from mineral estimates and remains robust in contrast-enhanced exams.
Which skeletal sites are supported?
Initial focus: trabecular vertebrae (L1–L4) and proximal femur; correlations also across cervicothoracic levels.
How are results delivered?
Structured DICOM SR and PDF including site-specific BMD (mg/cm³), guideline categories, QC flags, and optional FHIR/HL7 for EHR integration.
Request a demo
Tell us about your CT fleet and current workflow. We'll propose the fastest path to opportunistic QCT at your site.