skip to Main Content

An artificial intelligence system for automatic diagnosis using x-ray images

Www.innovus.co 6

Computer vision is a subfield of artificial intelligence that uses state of art deep learning architectures to predict features that may occur on images. In our case, we aim to learn anomalies, diseases and infections that are visible of medical images. Our initial dataset resonates from the dentistry domain. Here we aim to diagnose the problems that occur in oral radiology, with the vision spanning into all types of medical imaging eventually.

The system will accept an X-ray are an input, processes the image through a variety of different predictive models, and display the resulting diagnosis to the clinic doctor. Patients with severe cases will be marked and given propriety to treatment. The tool will be accessible as a secure cloud-based web application.

Target Industries

  • Public health:  Making an impact on public health, includes deploying the solution in public hospitals and tertiary education facilities. We would like to research and set up social impact bond for returns.
  • Private research facilities as well as private (dental) practices.
  • Insurance companies could also find the anonymized data useful for research.

Unique Features/Benefits

Instantaneous diagnosis for over 40 anomalies, and many different infections and diseases.

The same technology can be used in other medical imaging domains, such as cardiology and MRIs, among others.

Principal Researchers

  • James Faure – Industrial Engineering Stellenbosch University
  • Dr Shoayeb Shaik – Dentistry
  • Professor Andries Engelbrecht – Artificial Intelligence and Engineering

Innovation Status: Know how, Copyright

Fund Requirements: Seeking funding for product development and testing

Available for licensing: Yes

Share This

SU Logo 06

Related links

Resources

Contact US

Tel: +27 21 808 3826
Email: info@innovus.co.za

15 de Beer Street
Stellenbosch
7600

Campus Map

Social Media

Sign up to our newsletter

Sign up
Back To Top