Publications

Advancing Radiology
Through Research

Explore research and clinical studies evaluating artificial intelligence for lumbar spine MRI analysis, automated image assessment, diagnostic consistency and radiology workflow efficiency.

CoLumbo Significantly Reduces Lumbar Spine MRI Reporting Time
Clinical Study

CoLumbo Significantly Reduces Lumbar Spine MRI Reporting Time

236 lumbar MRIs5 experienced radiologists

This preliminary multi-center study evaluated the impact of CoLumbo on lumbar spine MRI interpretation and reporting efficiency. AI assistance significantly reduced interpretation and report preparation time across experienced radiologists.

Read publication →

Key points

236 lumbar MRIsRandomly selected examinations evaluated
52% time savingCoLumbo-assisted interpretation vs. unassisted reading
Multi-center study5 experienced radiologists across 3 European centers
AI-generated reportsEditable preliminary reports supported radiologist workflow
Lumbar spine MRI research on automated detection of degenerative changes
Research Publication

Automated Detection of Multiple Degenerative Changes in Lumbar Spine MRI

146 patientsT2-weighted MRI

This study evaluated CoLumbo's CNN-based approach for automated identification of vertebrae, intervertebral discs and multiple degenerative findings on T2-weighted lumbar MRI.

Read publication → View source ↗

Key points

146 patientsT2-weighted lumbar MRI
Multiple findings & structuresDegenerative findings across lumbar structures
Automated CoLumbo analysisCNN-based detection of degenerative findings
Expert reference standardCoLumbo compared with expert radiologist
Lumbar spine MRI study evaluating CoLumbo for central canal stenosis
Clinical Study

Prospective Evaluation of CoLumbo for Central Canal Stenosis

382 patients3 medical centers

This prospective multicentre study compared CoLumbo-assisted reading, unassisted radiologist assessment and CoLumbo alone for central lumbar spinal stenosis, evaluating diagnostic performance and observer variability.

Read publication → View source ↗

Key points

382 patientsProspective multicentre study
92.7% / 99.0%CoLumbo sensitivity / specificity for central stenosis
CoLumbo-assisted reading performed bestRadiologists using CoLumbo achieved the strongest results
Reduced observer variabilityEvidence supporting more consistent central stenosis assessment
CoLumbo evaluation of degenerative lumbar spine pathologies on MRI
Clinical Study

CoLumbo Evaluation of Degenerative Lumbar Spine Pathologies

90 patientsLumbar MRI

This preliminary study compared CoLumbo segmentation and detection results with the assessments of a senior neuroradiologist in patients with degenerative lumbar disease.

Read publication → View source ↗

Key points

90 patientsDegenerative lumbar pathology cohort
Expert comparisonCoLumbo compared with a senior neuroradiologist
Significant correlationCoLumbo grading correlated with expert assessment across lumbar levels
Up to 81.1% agreementHighest agreement achieved at L1–L2
AI-assisted lumbar spine MRI interpretation research
Time-Saving Evidence

AI Assistance Reduces Lumbar MRI Interpretation Time

51 MRI studiesWorkflow efficiency

Research from Sengkang General Hospital evaluated CoLumbo-assisted lumbar spine MRI interpretation, demonstrating shorter interpretation times and improved interobserver agreement among radiology residents.

Read article → View source ↗

Key points

CoLumbo reduced interpretation time23 → 9 min average interpretation time
Improved interobserver agreementMore consistent assessments with AI assistance
51 lumbar MRIsRandomly selected studies evaluated
Improved reporting efficiencySignificantly shorter interpretation time with AI assistance
Automatic segmentation of lumbar spine MRI using convolutional neural networks
Technical Research

Automatic Lumbar MRI Segmentation Using an Ensemble of 2D CNNs

Dice score 0.92MICCAI IVDM3Seg

Multiple 2D convolutional neural networks were combined into a 3D ensemble for automated lumbar MRI segmentation across different imaging planes and acquisition protocols.

Read publication → View source ↗

Key points

2D CNN ensembleMultiple models combined for automated segmentation
Dice score 0.92Reported segmentation performance
MICCAI benchmark ExplanationIVDM3Seg lumbar MRI segmentation benchmark
Multi-plane analysisDesigned to support analysis across different imaging planes and acquisition protocols
Corporate Document

Modern Slavery Policy and Annual Statement

Smart Soft Healthcare's policy sets out the company's approach to identifying, preventing and mitigating modern slavery, human trafficking, forced labour and child labour risks.

Read document →

Driving Innovation Through Collaboration

We collaborate with radiologists, researchers and healthcare institutions to advance AI in medical imaging.

Contact Us