Product:

ChestLink, ChestEye

ChestLink V1.0, ChestEye V2.4

Demo

Company HQ: 

Vilnius, Lithuania

Download Product Profile

Last Updated:

July 26th, 2021

ChestLink for normal reporting automation, ChestEye for prioritisation.

ChestEye and ChestLink are diagnostic aid tools and triaging solutions. The goal of these products is to save time for radiologists by identifying chest-x-ray images that have no abnormalities . ChestEye and ChestLink are based on the same AI algorithm but use different threshold scores.

ChestEye could prioritize the cases with abnormal images, including those with a high likelihood of having TB and also provide preliminary reports for them.

ChestLink also provides diagnostic aid by automatically issuing up to 25% of reports for cases without abnormalities on the chest x-ray. As a result, radiologists would not need to spend any time on these cases.

Contact info@oxipit.ai for demo site

Certification

ChestEye: CE Class IIa
ChestLink: CE Class II (in process)

Development Stage

On the market

Deployment

Both online and offline deployment available.

Intended Age Group

18+ years

Target Setting

Primary health centres, General hospital (above primary level), Teleradiology companies, Government/public sector, e.g. national TB program, Private sector

Current Market

Europe, South America, Australia, Asia, Middle East and Africa

Input

The product can be used to read images from any kind of CXR machine.
Format: JPEG, PNG, DICOM
CXR type: ChestEye: Posterior-anterior CXR and anterior-posterior CXR, ChestLink: Posterior-anterior CXR only

Output

Output includes:

  • Heat map

  • Dichotomous output indicating for each abnormality whether this is present or absent

  • Dichotomous output only indicating whether TB is likely present or likely absent

  • Location of each abnormality

The cutoff/threshold score can be adjusted / optimized for each client. Results are provided in a structured report.


Lung abnormalities detected include:

Abscess, Airfluid level, Atelectasis, Blunted costophrenic angle, Bronchiectasis, Calcification, Cavity, Consolidation, Fibrosis, Hyperinflation, Interstitial markings, Loculated pleural effusion, Lymphadenopathy, Mass, Nodule, Opacity, Pleural effusion, Prominence in hilar region,Pneumothorax

Tracheal shift


Lung abnormalities included in TB score:

Calcification, Cavity, Consolidation, Fibrosis, Lymphadenopathy, Opacity, Pleural effusion, Prominence in hilar region

ai4hlthlogoNEWtoUse.png

Hardware

Hardware requirements:
● 16GB of RAM
● 64bit CPU with 8 logical cores with AVX instruction support
● 1TB of disk space

Validation

X-ray machine validation is not mandatory, but preferable. Validation is possible with between 1,000 - 10,000 X-ray images.

Hardware

Hardware requirements:
● 16GB of RAM
● 64bit CPU with 8 logical cores with AVX instruction support
● 1TB of disk space

Server

Server requirements are the same as the hardware requirements

Integration

Software

Linux (preferably Ubuntu 18.04 LTS) with inbound SSH access

Processing Time

Up to 10 seconds

Data Sharing & Privacy

Server location (for online product)
Servers could be located by the need of the client

Data shared with manufacturer?
Whether data is shared with the manufacturer depends on agreement with the client.

De-identification (option to deidentify?)
Yes, additional software installation is required

Price

Please contact info@oxipit.ai for more information on price.

Oxipit software integrates with PACS and/or RIS systems via either DICOM or HL7 protocols. Images are automatically routed to Oxipit server and after analyses, Oxipit provides structural reports with priority back to PACS/RIS.

Software Updates

Frequency: Monthly
Cost: All costs are included in the pricing

Product Development Method

Supervised deep learning (CNN, RNN)

Training

1,000,000 CXRs from both non-public and open source datasets from Europe and Asia and South and North America.

Reference Standard

Majority Human Reader (Multi-read), also some examples with CT

Publications

Peer-reviewed publications are not yet available.

Does this page require updates? Send us a message:

Designed by Tasneem Naheyan

This website works best with browsers other than Internet Explorer.