Huge impact on diagnosis if paper is proven true

BEIJING — Chinese internet conglomerate Alibaba Group Holding has submitted a paper that claims it has successfully developed cardiovascular recognition technology, which if proven true, will have wide-ranging impact on the diagnosis and treatment of heart disease.

The group’s science and innovative technology research institute, Machine Intelligence laboratory of Alibaba DAMO Academy, submitted the paper to the International Conference on Medical Image Computing and Computer Assisted Intervention ahead of the 22nd convention in autumn.

According to a project leader at the Alibaba DAMO Academy, the lab used hundreds of thousands of samples to develop artificial intelligence that can now extract images of coronary arteries autonomously. It takes the technology just half a second to extract one coronary artery and 20 seconds or less to extract all coronary arteries, nearly 100 times faster than the conventional method which requires physical operations.

Alibaba DAMO Academy said it used a 3D convolutional neural network to explore images repeatedly and to detect all blood vessels. In order to diagnose coronary heart disease, coronary artery centerlines must be extracted accurately from computerized tomography angiography images. This is the most time-consuming part of the diagnostic process.

Coronary arteries have complicated geometrical characteristics and are very thin. Furthermore, it is difficult to distinguish between them and coronary veins. If blood vessels are partly clogged, it becomes difficult to extract all of them.

This Alibaba technology will take the burden of diagnosis partly off surgeons in cases of emergency when decisions have to be made quickly. The technology allows surgeons to discover disease sites quickly based on automatically created images.

Many “meditech” companies have been trying to develop AI that can help to diagnose diseases but cardiovascular ailments are complicated and there is not much existing data regarding image recognition.

In addition, as the heart beats without interruption, companies have found it difficult to reconstruct 3D images from scanned images. Also presenting major barriers for AI are the complicated meshlike structure of coronary arteries and the big differences in symptoms among individuals.

Alibaba DAMO Academy also claims that it has so far diagnosed 10 million patients using an AI-based pulmonary nodule detection technology. It is also developing liver cancer diagnosing technology.

36Kr, a Chinese tech news portal founded in Beijing in 2010, has more than 150 million readers worldwide. Nikkei announced a partnership with 36Kr on May 22, 2019.

Source: https://asia.nikkei.com/Business/China-tech/Alibaba-says-it-has-developed-AI-to-help-detect-heart-disease