top of page
Last Updated:
28 July 2021
This is a deep learning-based computer-aided detection model applicable to chest X-ray interpretation and intended to rule out tuberculosis (TB) in low prevalence populations such as contacts of TB index patients. This would be particularly useful to speed up the cascade of care for populations eligible for TB preventive treatment.
Certification
Not available
Development Stage
Under development
Deployment
Online and device-centric workflows
Intended Age Group
18+ years
Target Setting
Primary health centres, government/public sector, e.g. national TB programme
Current Market
Planned market: Brazil, Russia, India, China, and South Africa, other South American countries
Input
»» JPEG, DICOM; currently using JPEG but can easily support DICOM
»» Posterior-anterior or anterior-posterior chest X-ray
Output
»» Abnormality score for TB and for each covered finding. The location of abnormalities by side (left and right) and thirds (first, second or third).
1/1
Hardware
At the current stage, a notebook computer is necessary.
Validation
To be determined
Hardware
At the current stage, a notebook computer is necessary.
Server
To be determined
Integration
Not possible to integrate the current version with the client's legacy picture archiving and communication system (PACS).
Software
To be determined
Processing Time
To be determined
Data Sharing & Privacy
To be determined
Price
Open-source (free) software and artificial intelligence models distributed under the Affero GNU Public License version 3 (AGPLv3)
Software Updates
To be determined
Product Development Method
Supervised deep learning (CNN)
Training
2000 adult chest X-rays from Brazil, Canada, Benin, Indonesia, USA, India, China (the latter 3 from public databanks)
Reference Standard
Human reader
Publications
Peer-reviewed publications are not yet available.
This website works best with browsers other than Internet Explorer.
bottom of page