Cyted and Microsoft Research Collaboration Published in Nature Communications

• Cyted and Microsoft Research collaborated to build novel AI models for efficient screening of cancer cells that could reduce pathologist workload by as much as 63%
• Nature Communications published an article detailing the research

Cambridge based health company Cyted announced a publication in Nature Communications on research conducted in collaboration with Microsoft Research developing machine learning tools to enable earlier and faster detection of Barrett’s esophagus, a precursor to cancer.

“We have been delighted to collaborate with Microsoft Research to push the boundaries of what’s possible in medical imaging and screening technologies, creating optimal efficiencies from start to finish of the testing process” said Dr Marcel Gehrung, CEO and Co-founder at Cyted. “As we continue to expand, machine-learning will be an important tool to support our goals and make early cancer detection more accessible."

Esophageal cancer is one of the deadliest cancers with a five-year survival rate of less than 15%. Detecting Barrett’s esophagus, a precancerous condition, helps identify patients who are at a higher risk of developing esophageal cancer. Monitoring patients with this condition offers the best chance of early cancer detection and therefore earlier treatment of esophageal cancer, dramatically improving survival rates to 9 in 10 people surviving after 5 years.

The article titled “Enabling large-scale screening of Barrett’s esophagus using weakly supervised deep learning in histopathology” is an example of the innovative research that Cyted and Microsoft Research are advancing to optimize clinical workflows and improve the capabilities of earlier identification of cancer with technology.

Cyted’s capsule sponge technology, EndoSign® is a non-endoscopic test that collects cells from the esophagus in a process that takes less than ten minutes. Analysis of these cells is conducted by pathologists to identify biomarkers - a measurable biological indicator of the likelihood of a cancer developing.

Artificial Intelligence is being used to optimize the screening process by leveraging machine learning to help triage the large sets of pathology images. This saves precious pathologist time by screening out negative cases and controlling the quality of digitalized samples, cutting the pathologist’s workload down by as much as 63%.

“Microsoft Research seeks to advance science and technology to benefit humanity, often by identifying new applications for leveraging AI and machine learning,” said Javier Alvarez-Valle, Senior Director of Biomedical Imaging at Microsoft Research. “Collaborating with Cyted and sharing our work via open source will empower researchers and clinicians around the world to leverage this technology in their fight against cancer.”

The full article in Nature Communications can be viewed here.

A blog post written by Microsoft Research can also be found here.