Every year, around 1.3 billion people have to take medical image films when visiting a doctor in China. While how many experts are there in the authority tertiary grade-A hospitals? How many people's problems on the films are discovered by these experts?

"To ensure everyone can enjoy quality medical treatment is PereDoc's mission," said PereDoc CEO Lian Jing at the strategy briefing of intelligent medical imaging products on December 19.

The PereDoc product-launch event gathers heavyweights.

The wisdom results that break world records

According to statistics, the medical image data grows by 30% yearly in China, but among radiologists it grows by just 4.1%. Moreover, the worsening shortage of radiologists is one of the important reasons that causes the 40% misdiagnosis or missed diagnosis of cancer in early detection.

"What we'd like to do is to fill up the 40% misdiagnosis or missed diagnosis with AI," said Lian Jing.  

CEO Lian Jing releases five new products of PereDoc

"The accuracy of PereDoc's assisted diagnostic products is almost comparable to their American counterparts," Yu Sicheng, former vice president of Alibaba and CEO of iTrigger, said cheerfully. "And some even surpassed their competitors."

The detection rate of PereDoc pulmonary nodule imaging diagnosis system 3.0 is 94.9％ for 2-5mm pulmonary nodules and 99.2％ for 5mm pulmonary nodules, breaking the world record of 89.7% created by Alibaba Cloud ET in the international authoritative pulmonary nodule testing contest LUNA16 and becoming the world's No. 1.

At the conference of December 19, PereDoc launched five assisted diagnostic products, of which pathological assisted screening of lung cancer and assisted diagnosis of space-occupying liver lesions are first launched in China and the other three take the lead in the world.

The first pathological assisted screening system of lung cancer based on in-depth learning has the epoch-making significance. At present, China's Department of Pathology lacks 40,000 doctors and nearly 40% operations have not been analyzed by pathologic slices, resulting in frequent missed diagnosis and misdiagnosis. The pathological assisted screening system of lung cancer 1.0 launched by PereDoc can load digital case scan results of 200,000*200,000 pixels, automatically detect lung cancer cells, greatly improve the work efficiency of pathologists and make more pathological diagnosis possible.

Aided diagnosis of space-occupying liver lesions is also launched in China for the first time. It can automatically detect and select lesions and analyze the halo sign and mosaic sign of lesions so as to help clinicians to a large extent, guide the treatment and speculate prognosis.

Besides, the aided-diagnosis system for pulmonary diseases has realized the overall coverage of major pulmonary diseases. It can identify five abnormal symptoms and diagnose over 100 pulmonary diseases, thus playing an important guiding role clinically. Another new product, the assisted screening system of chest x-rays covers more than 20 symptoms of major cardiothoracic diseases, such as nodule mass shadows, patchy shadows, cavity shadows, pneumothorax, calcifications and fracture. As the most common and widespread preliminary screening and testing method of pulmonary diseases in the hospital or the medical center, x-rays along with the assisted screening system will largely reduce doctors' the work pressure. Mammography aided molybdenum target screening system 2.0 is further upgraded on the basis of the detection rate of micro-calcification. Mammography technology has been very popular and become an indispensable diagnostic method of breast cancer screening for females aged above 40. Mammography aided molybdenum target screening system has greatly advanced the rapidity and veracity of breast cancer and won the unanimous praise of clinicians.

The original Gold Standard annotating mode

Peter Thiel, an angel investor of Facebook, stated in his work Zero to One that happy families are all alike but every unhappy family is unhappy in its own way. Yet it is rather the reverse situation in the business. The reasons for the success of enterprises are various that all monopoly enterprises win its monopoly by solving a unique problem. The reasons for their failure are the same that they cannot stay alone from the competition.

"The original gold standard annotating mode explains our 99.2% accuracy rate," said Lian Jing.

According to Wu Chongchong, chief physician at Chinese PLA General Hospital, "Unlike regular annotation that only labels the location and size of nodules and lung window morphology, PereDoc's gold standard annotating mode also labels the nodule or the surrounding situation, intrapulmonary metastasis, enhanced characteristics of CT values and features of regional lymph nodes and bone destruction based on the pathological results, so as to obtain more accurate diagnosis results and help doctors to screen out suspected pulmonary nodule lesions quickly."

The main problem for AI companies in the medical sector is that the current medical imaging scarcely labeled the lesions. However, such systematic data preparation is highly specialized and therefore requires the close cooperation of professional doctors. This is also the uniqueness of the medical industry.

As the medical data have not realized interconnection, the medical imaging data application in China is still in its infancy. "We have data, a strong expert team. That is our edge." Lian Jing said confidently.

AI is a good helper for clinicians

"Big data is called the crude oil in the era of AI." Professor Wang Rengui, of Shijitan Hospital of Capital Medical University, has more than 30 years of clinical experience and is familiar with AI technology. He kindly referred to AI (abbreviation of artificial intelligence) as "love" (AI being pronounced ai in Chinese, meaning "love"). He said, "More than 80% of the medical data comes from the medical imaging. The massive big data in the medical sector achieve the great revival of 'AI China'."

As a clinician, Wang Rengui began with the earliest assisted diagnostic product of PereDoc, "In the past few months, I have lectured on the user experience three times. But I have to reorganize my thinking for each lecture, as the product upgrade is so fast! Within a few months, we witnessed 1.0, 2.0 and 3.0. I think it won't be long from 30.0. "

Wang Rengui said he could read 30 patients' films a day. Each patient's film has a huge amount of data, as much as 8G. "All the work is done by manpower. By afternoon, the head becomes muddled and the temper gets shorter. It's quite difficult to keep a sober mind and deal with the complicated imaging data. "

"But, with the help of AI it's a piece of cake!" said Wang Rengui with a smile.

Recently, the clinical trial of intelligent diagnosis conducted by the Harvard Medical School indicated that, with AI-assisted breast-cancer diagnosis, the problem of misdiagnosis was reduced to 0.5% from the original 4%.

"AI is clearly efficient, so doctors should have the right attitude about it," said Wang Rengui. "As human beings we aren't competing with AI, but instead we should regard it as an effective helper." He explained that the development of humankind and machines are on different paths. Consequently, AI will replace a large number of positions that people don't like or are unable to perform.

Closing the gap between diagnosis and treatment for doctors and hospitals

"I feel that my students have surpassed me," said Professor Wu Jianlin, chairman of the Cardiothoracic Team of the Chinese Medical Association.

Not long ago, a student of Professor Wu found a 3mm pulmonary nodule and wrote a highly detailed image report. This surprised Wu, who had more than 30 years of clinical experience. The student, however, admitted it was the PereDoc image-aided diagnosis platform that helped him complete this incredible task to such a degree of accuracy and professionalism.

"The most important thing is to elevate and make more uniform the diagnosis and treatment level of doctors with different qualifications and from different areas," said Professor Zhao Shaohong, of the Radiology Department of Chinese PLA General Hospital. That facility is one of first tertiary grade-A hospitals to use PereDoc assisted diagnostic products. "Just as Lian Jing said, I think the emergence of AI will give more people access to quality healthcare."

The problem of uneven distribution of medical resources is chronic in China. Good teachers are scarce, and they tend to gather in tertiary grade-A hospitals. Few of them are willing to work at grassroots medical institutions.

"With the help of AI-assisted diagnostic equipment, the gap between diagnosis and treatment to which doctors have been subject can be bridged," said Professor Zhao Shaohong with a tone of optimism. "The inequality of medical resources that has been criticized for years will be corrected."

The professor and his colleagues, having used PereDoc assisted diagnostic products for several months, began to consider undertaking more film-reading business from grassroots hospitals and directly outputting the technology and resources of top-level hospitals to grassroots ones. They believed it would be possible to elevate their respective levels of diagnosis and treatment.  

Full devotion

A recent report published by Stanford University--titled AI index 2017 annual report--pointed to the fact that because there is no reasonable data with which to analyze the status of AI, to date most AI projects have blindly followed suit.

Yu Sicheng, formerly a vice president of Alibaba and the CEO of iTrigger, offered his opinion on the matter. "People need to stop and think seriously about whether AI is in rapid development or will become a capital bubble. Maybe there will be some profound changes in the medical AI market during the next few years. Some companies will be merged, and only those companies that succeed in the development of talented teams, the accurate capture of medical needs, the improvement of product technology and growth in market share can survive and prosper."

The guests engage in an intense discussion of the future of intelligent medical treatment at the PereDoc product-launch event.

State-owned "tycoon enterprises" have few forward-looking plans for AI, and most of them have a wait-and-see attitude. "Some disdain it, some tease it, and some have attempted to portray themselves as rationalists. However, few companies to give full attention the way that PereDoc does. Devoted efforts have given BAT (Baidu, Alibaba, and Tencent) an enviable profile, and I believe that in a few years there will be AI tycoons to give this technology their full attention."

"The peregrine falcon is the world's fastest flying raptor and also the one with the keenest eyesight." Lian Jing said. "This is the origin of our name PereDoc. We can capture the smallest lesions accurately and quickly so that everyone can have access to high-quality medical treatment."