AI diagnostic tool that can identify COVID-19 characteristics in CT scans
As COVID-19 spreads exponentially, CT imaging departments are under significant pressure to optimize their performance levels and understand the virus's intricacies. But the number of suspected cases is growing at an often-exponential rate around the world and clinicians are struggling to keep pace with demand.
A CT machine typically produces 300 to 400 chest images per patient per scan and it takes between five and 15 minutes for a doctor to analyze this information. A patient with COVID-19 also receives multiple scans with, on average, more than 2,000 CT images during the entire hospitalization process. Manually reading and comparing this imaging data for every patient requires a huge amount of effort in optimal conditions. When teams are nearing capacity, however, the individuals assessing these scans are often fatigued and operating under excessive workloads; human error could easily become an issue.
When precision and pace are required, CT Image Analytics for COVID-19 can help, significantly improving the testing accuracy and detection efficiency rates. Our CT Image Analytics for COVID-19 technology can assist in identifying characteristics of coronavirus pneumonia in CT scans with about 96% accuracy. It is 60 times faster than human detection methods, taking less than four seconds to run each test and transmit the data. More than 160 public institutions in China are currently using this technology. As of March 14, 2020, the system has already analyzed more than 240,000 CT image volumes (around 13,000 per day on average). This technology relies on cutting-edge deep learning algorithms, which have been trained using 5,000 cases to understand the differences between COVID-19 pneumonia, common pneumonia, and other conditions. This advanced technology can also estimate the proportion of pulmonary lesions for those affected by COVID-19.
The CT Image Analysis for COVID-19 technology benefits from a powerful combination of scalable cloud-based services and accurate deep learning algorithms. The AI system identifies the virus through computed tomography scans of the chest. The algorithm has been trained with data and CT scans from more than 5,000 confirmed coronavirus cases so far and taps into deep learning to study patterns of infection. The algorithm learns the differences between COVID-19 pneumonia, common pneumonia, and other situations. The technology can then predict the probability of COVID-19 pneumonia and common pneumonia based on the input CT images.
This CT Image Analysis technology can also output the lesion masks and affected lung volume ratio, helping doctors to effectively measure the development or treatment of COVID-19 patients. Thanks to the cloud-based technology it relies on, CT Image Analytics for COVID-19 is highly scalable, high-speed, and seamlessly exchanges image and case data across different medical systems. This results in an improved response time, which is important when working as part of a global community to lessen the impact of COVID-19. Alibaba’s API connects these cloud-based technologies and the DAMO algorithm, linking the Alibaba Cloud technologies to the hospital’s PACS (Picture Archiving and Communication Systems).
CT Image Analytics for COVID-19 is easy to deploy. Using one of two methods, medical institutions can:
Once deployed, medical professionals can benefit from a rapid and highly accurate analysis tool, helping them understand the nuances of this disease for every case, avoid errors and keep pace with the pandemic.