Infervision Using Deep Learning to Detect Cancers

Infervision Using Deep Learning to Detect Cancers
Share this:

For some time now, human beings have been relying on a doctor’s trained eye in diagnosing illnesses from medical images.

Infervision, a Beijing-based company, is among several artificial intelligence (AI) startups across the globe that are currently on track to boost imaging analysis by leveraging deep learning technology.

To date, the startup has fundraised about $70 million from top investors such as Sequoia Capital China.

Infervision started by detecting cancerous lung cells, which are among China’s common cause of death.

Recently, the company revealed its plans to extend its computer vision capabilities to chest-based conditions such as cardiac calcification.

The announcement took place at the annual Radiological Society of North America’s conference, which was held in Chicago.

“By adding more scenarios under which our AI works, we are able to offer more help to doctors,” said Chen Kuan, the chief executive officer and founder of Infervision.

While a physician can identify dozens of illnesses from one image scan, artificial intelligence requires to be trained how to detect numerous target objects in a single trial.

However, Chen is convinced that machines already outdo human beings in other areas.

For starters, they can read faster compared to humans.

Normally, doctors take about 15-20 minutes in scrutinizing a single image, while Infervision’s artificial intelligence cannot only synthesize the visuals but also compile a report in less than 30 seconds.

Artificial intelligence (AI) helps in dealing with the misdiagnosis issue.

Medical Weekly, a Chinese clinical newspaper, recently reported that physicians with under five year’s experience only obtained their responses right 44% of the cases when detecting black lung, an illness that is common amongst coal miners.

“Doctors work long hours and are constantly under tremendous stress, which can lead to errors,” said Chen.

The founder said that Infervision could boost the accuracy level by 20%.

Artificial intelligence (AI) has the potential to fill in for physicians in remote hinterlands whereby healthcare service falls short, which is a common case in China.

Like any other similar company, Infervision has to constantly train its algorithms using data from varied sources.

As of recently, the company is collaborating with about 280 hospitals, whereby 20 of them are located outside China.

Aside from that, Infervision steadily adds 12 new partners each week.

Chen, a Shenzhen-based native, established Infervision after dropping out from the University of Chicago where he was pursuing a doctoral program.

In the initial six months of Chen’s entrepreneurial journey, he visited about 40 hospitals around China without any success.

“Medical AI was still a novelty then. Hospitals are by nature conservative because they have to protect patients, which make them reluctant to partner with outsiders,” Chen recalled.

Finally, the Sichuan Provincial People’s Hospital gave him a chance.

With two other founding members, Chen obtained a small volume of image data, shifted into an apartment near the hospital, and started the company.

“We observed how doctors work, explained to them how AI works, listened to their complaints, and iterated our product,” said Chen. Infervision’s product proved itself, and soon its name became popular among healthcare experts.

“Hospitals are risk-averse, but as soon as one of them likes us, it goes out to spread the word, and other hospitals will soon find us. The medical industry is very tight-knit,” said the founder.

Infervision has had challenges in its abroad markets.

In the US, for instance, the company is required to visit physicians only after making appointments.

Chen also confessed that most US-based hospitals were not convinced that a Chinese startup could offer advanced technology, but went ahead to welcome Infervision after realizing the company’s potential and abilities.

Share this:

Leave a Reply

avatar
  Subscribe  
Notify of