Last Updated:
23 January 2025
Genki (‘Healthy’ in Japanese) is DeepTek’s AI-powered public health screening solution. Genki is a first CADe solution that uses US FDA cleared Chest X-Ray AI models covering entire lung as an organ for identification of Tuberculosis, Nodules and 13 other pathologies.
Genki provides an end-to-end solution with a Patient Registration and Vulnerability Index Management Solution, Offline/Edge solution and Online- Hub solution with comprehensive analytics. Genki has undergone independent third-party evaluation by independent firms like Friends for International Tuberculosis Relief (FIT) where it meets the minimum requirement for a community-based triage test recommended by WHO with Sensitivity of 91% and Specificity of 89%. Genki has also helped increase the yield of finding TB patients by 25 times i.e. from identifying 20 to 500 suspects per 100,000 screened.
Certification
US FDA 510(k) clearance
Singapore HSA
Indian CDSCO
Thai FDA
Malaysia FDA
Indonesia FDA
Kenya Poison Board
CE marking (Applied)
Development Stage
On the Market
Deployment
Online & Offline
Intended Age Group
4+ years
Target Setting
Genki can be used in communities for active case finding by government bodies or National TB Programs. The solution can also be deployed by primary or tertiary care centres for triaging of TB suspects. The AI is also used in Public /Private hospital as a Lung Health Screening tool,
Current Market
India, Philippines, Thailand, Mongolia, Bhutan, Singapore, Indonesia, Nigeria, Congo, Cameroon etc
Input
Can be used to read images from any chest X-ray machine and model
Chest X-ray image format: DICOM, JPG, PNG
Chest X-ray type: Posterior-anterior chest X-ray, Anterior-posterior chest X-ray
Output
Output includes:
Heatmap
Dichotomous output indicating whether TB is likely present or absent
Dichotomous output indicating whether each abnormality is likely present or absent
Probability score for TB
Probability score for each abnormality
Location of each abnormality
Default threshold for TB is 0.5 and can be adjusted.
A well formated tabular pdf report is created based on Genki findings for the screened person.
Lung abnormalities included in the TB Score: Calcification, Cavity, Consolidation, Fibrosis, Opacity, Pleural effusion, Infiltrations
Additional findings reported by the product: Atelectasis, Bronchiectasis, Calcification, Cavity, Fibrosis, Mass, Nodule, Opacity, Pleural effusion, Pneumothorax, Pleural thickening, Pneumonia
Hardware
A Windows Pro/Linux with 16 GB RAM, 8 core, i5 11th Generation, 2,5 GHz, 1 TB HDD is required for Genki.
Server
To install Genki on a cloud based server or On prem server with server specification matching to 16 GB RAM, 1 TB Disk, 8 Core, 2.5 GHz Windows or Ubuntu OS.
Integration with X-ray Systems
Integration with PACS and Legacy Systems
Genki can be integrated to work with any X-ray manufacturer that supports DICOM protocol (Ultraportable, Portable, Mobile, Fixed X-ray machines). X-ray image is transferred to Genki over DICOM protocl using the AE title configuration.
Yes, Genki can be integrated with PACS and Legacydata system as long as to be integrated system supports DICOM protocol.
Software
Windows/ Linux OS, Docker, Web Browser is required.
Processing Time
Less than 30 seconds for TB findings.
Data Sharing & Privacy
Patient data is always protected wih Genki under the HIPAA compliances. Data is not shared with the developers during the product usage at any point. Genki can be configured to anonymise the data prior to processing to safeguard the PHI for it on-premise or on-cloud deployment.
Software Updates
Genki undergoes a 6-8 month updates cycle to insure product performance.
Product Development Method
Product development method follows standard SDLC lifecycle with AI model development lifecycle. Various Deep Learning algorithms are utilised to process the X-ray image and provide the findings for pathologies.
Training
Genki has been trained on over a million CXRs images inclusive of data from different geographies
Reference Standard
Ground truth/reference standard created by human readers (Radiologists) and GeneXpert/smear data was used to train the Genki.
Publications
Please visit https://www.deeptek.ai/publications
This website works best with browsers other than Internet Explorer.