JVIEWER-X has dual-purpose: firstly, it can be used as a triage and prioritization tool for radiologists or teleradiologists. Secondly, it can be used to as an automated chest X-ray interpretation tool that can detect several pulmonary abnormalities, including signs of tuberculosis (TB) and others: atelectasis, consolidation, fibrosis, mass, nodule, pleural effusion, pneumothorax, pneumonia, cardiomegaly, effusion, infiltration, oedema, emphysema, fibrosis, hernia, and COVID-19.

Certification

CE-marked Class I, Australia TGA.
Others: Korea FDA, New Zealand MEDSAFE, Turkey MoH, Indonesia MoH, Thailand FDA

Development Stage

On the market

Deployment

Online & offline

Intended Age Group

10+ years

Target Setting

Primary health centres, general hospital (above primary level), teleradiology companies, government/public sector, e.g. national TB programme, private sector

Current Market

South Korea, USA, India, China, Japan, Indonesia, Laos, Thailand, Russia, Dubai, and Brazil

Input

»» Can be used to read images from any kind of chest X-ray machine (vendor neutral)

»» Chest X-ray image format: JPEG, PNG, DICOM

»» Chest X-ray type: posterior-anterior chest X-ray, anterior-posterior chest X-ray, portable

Output

A structured report consists of:

»» Probability score for each abnormailty as well as dichotomous output indicating whether the abnormality is present or absent

»» Probabilty score for TB as well as dichotomous output indicating whether TB is likely present or likely absent

»» Heat map  

Default threshold probability score: 50% but can be adjusted.

The product can detect the lung abnormalities: TB, atelectasis, consolidation, fibrosis, mass, nodule, pleural effusion, pneumothorax, pneumonia, cardiomegaly, effusion, infiltration, oedema, emphysema, fibrosis, hernia, and COVID-19. 

Hardware

Product does not need GPU (offline solution case). Intel 5 or higher generation CPU with Intel Graphics, full HD monitor, recommended 20 GB HDD or SDD. Compatible with laptop or mini PC.

Validation

Validation only required if chest X-ray images are from a low dose machine that affects the image quality.

Hardware

Product does not need GPU (offline solution case). Intel 5 or higher generation CPU with Intel Graphics, full HD monitor, recommended 20 GB HDD or SDD. Compatible with laptop or mini PC.

Server

Minimum 16 GB RAM, Ubuntu 16.04

Integration

It is possible to integrate the product with the client’s legacy picture archiving and communication system (PACS). Please contact the sales team at sales@jlkgroup.com.

Software

Windows 10, minimum 8 GB RAM

Processing Time

15-20 seconds (depending on the performance of the hardware)

Data Sharing & Privacy

»» Server location: the product can be set up in a public cloud such as Azure or Amazon Web Services

»» Data are automatically shared with the developer for the online solution but not for the offline solution. In the online solution (via cloud), the developer cannot access the X-ray database, and the data are automatically deleted after 24 hours.

»» There is an option to de-identify data. The X-ray data are anonymized before artificial intelligence analysis.

Price

»» Please contact the company's sales team at sales@jlkgroup.com for pricing information

»» There is a pricing difference between online and offline solutions

»» Please contact the company's sales team at sales@jlkgroup.com for details on upfront installation and set-up costs

Software Updates

»» Software is updated twice a year

»» Please contact the company's sales team at sales@jlkgroup.com for pricing information

Product Development Method

Supervised deep learning (CNN, DBNs)

Training

The product was trained on 1.5 million chest X-rays from Republic of Korea, Malaysia, India, Indonesia, China, and South Africa

Reference Standard

Culture, smear, GeneXpert, computed tomography (CT), and human reader

Publications

»» Kwon, T., Lee, S.P., Kim, D. et al. Diagnostic performance of artificial intelligence model for pneumonia from chest radiography. PLoS One. 2021;16(4):e0249399. https://doi.org/10.1371/journal.pone.0249399