Columbia Technology Ventures

Neural network that can diagnose SARS-CoV-2 infection using phlegm sample images

This technology is a machine learning approach to identify patients infected with coronavirus (SARS-CoV-2) using phase-contrast images.

Unmet Need: Rapid identification of coronavirus infection in patient samples

Current approaches to diagnose coronavirus infection are dependent on detection of the viral genome or viral antigens. These methods commonly require specialized equipment, may have low detection limits, and may not be adequately rapid. As such, an efficient and scalable approach for identifying coronavirus infection is needed to combat the epidemic.

The Technology: Machine learning for rapid and scalable diagnosis of coronavirus infection

This approach utilizes machine learning to determine whether a patient has been infected by coronavirus. The patient provides a phlegm sample, of which phase-contrast images are then taken. These images are processed by the algorithm, which is trained on infected and uninfected images, to determine if the patient’s phlegm sample is infected by coronavirus.

Applications:

  • Diagnosis of coronavirus infection
  • Diagnostics of other infections

Advantages:

  • Rapid diagnosis of coronavirus infection
  • Scalable
  • Cost-effective

Lead Inventor:

Helen Lu, Ph.D.

Tech Ventures Reference: