Nexus AI CXR

Nexus AI is a CE MDR Class IIb certified medical device designed to support clinical interpretation of chest X-rays. It analyses chest X-rays and indicates any findings that are not consistent with a normal lung field, including findings suggestive of tuberculosis (TB).

Nexus AI

Revolutionizing Lung Health with AI Support

Nexus AI CXR provides AI-assisted chest X-ray analysis in under 45 seconds, supporting trained healthcare professionals in screening, triage and lung health programmes.

CE MDR Class IIbISO 13485:2016Published in The LancetDeveloped with GoogleAvailable through PFSCM / Wambo
Nexus AI CXR workspace showing chest X-ray with AI heatmap
<45 sec
Analysis time per chest X-ray, from upload to AI findings
460 000+
Chest X-rays in initial training data, across 10 countries
0.897 AUC
Independent Stop TB evaluation, The Lancet Digital Health (2024)
The platform

Nexus AI Platform

Simple. Understandable. Built for real clinical environments.

An integrated enterprise AI platform supporting population screening, radiology workflows, and occupational lung health programmes - developed for flexible deployment across diverse clinical environments. Designed to support trained healthcare professionals. Clinical judgement always remains with the clinician.

  • Cloud or on-premises · Online or offline
  • DICOM, JPEG, PNG and TIFF inputs with automated quality verification
  • Real-time AI findings with isolines, heatmaps, and confidence scores
  • LLM-powered clinical insights and structured reports (Beta)
Nexus AI dynamic study queue
The solution · The expansions

Two AI Models. Run in Parallel.

The Nexus AI algorithm comprises two core models (Normal and TB models) that run in parallel. Multi-pathology and LLM capabilities build on the certified base and are advancing in Beta.

CE Certified

Normal model · FIR

Detects any deviation from a normal lung field. Isolines and heat maps for every finding. Findings are classified as “FIR” (further investigation required).

CE Certified

TB model · TB Suggestive

Identifies patterns suggestive of tuberculosis. Isolines and heat maps for TB-suggestive cases. Findings are classified as “TB Suggestive” with a confidence score (0-100).

Beta

Multi-Pathology classification

Adds descriptive labels for up to 10 radiological findings beyond TB - with isolines, heatmaps, and scores. Silicosis - external validation underway with Stop TB.

Beta

LLM Clinical Insights

Context-aware clinical summaries and structured reports. Clinicians add clinical context and the LLM assists interpretation. Internally validated, and we are currently in the process to expand our CE certification.

CE MDR Class IIb certified · Developed with Google · Independently validated

Key Benefits

Nexus AI CXR combines clinically validated performance, rapid triage and enterprise-scale deployment in one affordable, connected platform - built for high-volume lung health screening, TB triage, occupational health and radiology workflow optimisation across public and private healthcare systems.

01

Regulatory-ready confidence

CE MDR Class IIb certified medical AI software, supported by an ISO 13485:2016 quality management system - providing a strong foundation for clinical deployment.

02

Validated clinical performance

Independently evaluated by the Stop TB Partnership and published in The Lancet Digital Health. Nexus ranked among the top-performing products evaluated.

03

Results in under 45 seconds

Instant AI-based chest X-ray triage can reduce the initial screening turnaround from hours or days to under 45 seconds per study, supporting same-day decisions and faster referral.

04

Global Fund quality-approved

Assessed and quality-approved by the Global Fund, supporting deployment in donor-funded TB programmes.

05

Procurement and programme access

Available through PFSCM/Wambo. Nexus is also listed on the Stop TB AI4HLTH platform. WHO prequalification evaluation is in process.

06

Proven rollout globally

Deployed at scale across multiple countries, with strong demand and interest from Ministries of Health, funders and implementing partners.

07

Affordable by design

A commercially sustainable model supports continuity beyond funded project cycles - helping national programmes and implementing partners broaden access without sacrificing quality.

08

Long-term agreement with Google LLC

Developed in collaboration with Google under a Master Licence, Development & Commercialisation Agreement - enabling scalability, continued innovation and future AI development. LLM capabilities are already implemented in Beta.

09

Broader lung health ecosystem

Integrates with HCF, CloudX, clinical workflows, central radiology reporting and future digital health solutions, enabling clinical management, reporting, longitudinal EHR and programme analytics.

The outcome: Faster prioritisation, more consistent screening and a practical path to scale.

Evidence & performance

Performance you can Interrogate

Nexus AI CXR was developed on over 460,000 chest X-rays from 10 countries across diverse populations and imaging modalities. Both the Normal model and the TB model met or exceeded the World Health Organization’s performance thresholds (≥90% sensitivity, ≥70% specificity) in validation by Nexus and Google.

In an independent external evaluation of 12 commercially available CAD products by the Stop TB Partnership, published in The Lancet Digital Health, Nexus AI CXR achieved:

0.897
AUC (95% CI 0.872-0.922), statistically indistinguishable from the highest result
89.1%
Highest fixed-threshold sensitivity (0.5 operating threshold)
67.1%
Specificity at 90% sensitivity - CI reaching the 70% WHO target
Top 2
Among the top-performing products, confirmed across HIV-positive, female, male and age subgroups

Source: Qin ZZ, Van der Walt M, Moyo S, et al. The Lancet Digital Health, 2024, 6(9), e605-e613.

Stop TB Partnership CAD evaluation results, The Lancet Digital Health 2024

One Comprehensive Clinical Workspace

The Nexus AI workspace is built around real clinical workflows.

Upload

DICOM, JPEG or PNG

AI Analysis

< 45 sec · two models in parallel

Findings

FIR · TB Suggestive · scores

PDF Report

Structured & shareable

Linkage to care

Referral & follow-up in HCF

AI findings, scores & overlays

Real-time AI findings with isolines, heatmaps, and confidence scores (0-100). Clinicians switch between overlays.

Dynamic study queue

Live tracking and management of incoming X-ray studies. Filter by DOB, score, gender, upload date, or study ID.

Structured PDF reporting

Standardised PDF reports with prediction scores, visual overlays and findings summary - print, email, or archive to PACS.

DICOM, JPEG, PNG & TIFF inputs

Full DICOM AP/PA support with automated quality verification. Also accepts JPEG, PNG and TIFF. DICOM-web and WADO-RS APIs for PACS.

Flexible deployment

Deploy on cloud, on-premises, or fully offline. Same architecture, same regulatory compliance, same clinical performance everywhere.

Enterprise security & compliance

Role-based access (RBAC), JWT authentication, TLS encryption, AES-256 at rest, full audit logging. GDPR & ISO 27001 aligned.

Decision-support tool. Nexus AI CXR 1.0 is a CE MDR Class IIb certified medical device intended to assist qualified healthcare professionals. It does not replace diagnostic interpretation by a trained clinician. All outputs must be interpreted in conjunction with the patient’s clinical context, medical history, and additional diagnostic results. The system does not alter the original image. Multi-pathology classification (Nodule, Fibrosis, Silicosis, Lung cancer and others) is available as a Beta feature and is not part of the CE MDR Class IIb certification. Nexus AI CXR is developed by Nexus Intelligence (Pty) Ltd.

Ready to see Nexus AI in your workflow?

Book a platform walkthrough with our clinical team. We’ll show you live AI analysis, integration options with HCF and CloudX, and walk you through deployment in your environment.

Request a Demo