SAMPL Clinical Research | Covid-19 Research

COVID-19 Diagnostics and Monitoring via Image-Based AI

We have put together a team with machine learning and data science specializations that are working together with our clinical forum to develop and implement image analysis techniques using AI that will help with identification, triage and diagnosis of COVID-19 patients or suspected carriers. These methods will also be used to monitor disease progression and prognosis as well as patients post-disease. We are also developing efficient pooling methods for PCR classification that will allow for a large amount of tests with reduced cost and time. Finally, we are exploring simple methods for COVID-19 detection such as voice and heart-rate via radar signals.

Our team includes members from the Weizmann Institute and volunteers from many data science based companies including Mobileye, Intel and Microsoft, and many hospitals throughout Israel. The 8400 health network is also supporting this activity. The work is done in order to benefit patients and hospitals and will be made public so that everyone can benefit from it.

Please contact us if you would like to take part in this effort.

The Challenge

• CT gives very clear picture of lungs and ground glass opacities which indicate disease
  • Expensive and non-portable equipment
  • Disinfecting the CT machine between patients is complicated and lengthy
• X-ray is more readily accessible and easier to clean
  • Findings not always very clear

• Ultrasound is portable and relatively easy to disinfect; however, currently not considered a viable lung diagnostic tool
  • Little research on US-based diagnosis of COVID-19

    

Goals

• Using AI to improve diagnostics and monitoring with X-ray
• Using AI on X-ray/CT to predict which patients will develop severe symptoms
• Develop tools to detect and monitor COVID-19 with ultrasound
• Monitoring recovered patients (X-ray and CT) and examining long term affects
• Exploring alternative simple methods for detection such as voice and heart-rate via radar.

Watch

The Team

The ultrasound work is done in collaboration with colleagues in Italy:
Libertario Demi, Federico Mento, Gino Soldati, Andrea Smargiassi, Riccardo Inchingolo, Elena Torri, Tiziano Perrone

Publications

• O. Frank et. al, "A Framework for Integrating Domain Knowledge into Deep Networks for Lung Ultrasound, and its Applications to COVID-19", IEEE Transaction on Medical Imaging, March 2021.

• D. Keidar et. al, “COVID-19 Classification of X-ray Images Using Deep Neural Networks”, European Radiology, pp. 1-10, May 2021.

• Shai Bagon et. al, “Assessment of COVID-19 in Lung Ultrasound by Combining Anatomy and Sonographic Artifacts using Deep Learning”, ASA, December 2020.
• Yishai Elyada et. al, “Point of care image analysis for COVID19”, Speech and Signal Processing (ICASSP), Virtual, June 2021.

Related Activities

• Co-Chair of the InterAcademy Partnership’s COVID‐19 Advisory Group
  (Link https://www.interacademies.org/node/52980)
• Task Force Leader in 8400 Health-Network on AI in the Service of COVID Imaging Data
  (Link https://www.8400thn.org/ctf1)
• Team Member in drafting a national Report By WIZO on Exit Strategies
  (Link http://www.wizo.org.il/page_35766)
• AI Cures Team Member
  (Link https://www.aicures.mit.edu/team)