AI in Medical Imaging in China: How to Switch from Adoption to more Sustainable Adoption

Medical imaging is a tool that can help doctors view what is happening inside a patient’s body, thereby assisting in diagnosing, monitoring and treating an ongoing illness for the patients. 

The first medical image was captured by Wilhelm Conrad Roentgen, a German professor in physics, on November 8, 1895. It is also when the X-ray was invented. Since then, other revolutionary medical imaging tools have also emerged as technology advances. Now common types of medical imaging include X-rays, CT (computed tomography) scans, MRI (magnetic resonance imaging), ultrasound, and nuclear medicine imaging.  

China has also seen significant adoption of medical imaging in hospitals. Especially in recent years, the implementation of AI has further advanced medical imaging technology and made it an even better helper for doctors to some extent. 

“In 2014, we had 100+ staff working in the radiology department, including radiologists and radiographers. The average workload per doctor per day at that time was 1422 patients. The staff number increased to 196 in 2020, a 66% growth. However, the workload increased to 3300 patients per doctor per day in 2020, a 133% growth. We have certainly seen the benefit of adopting AI-powered medical imaging in our hospital and how it has enabled us to see more patients efficiently.“ Dr. Hu Hongjie, Director of Department of Radiology in Run Run Shaw Hospital, Zhejiang University School of Medicine told EqualOcean. 

Adoption of AI-based medical imaging in China

The integration of artificial intelligence in medical imaging is able to bring benefits for both medical researchers and radiologists, especially in the aspect of better utilization of medical imaging data and reduction of medical error or misdiagnosis. 

Since the adoption of medical imaging tools, China and the globe has generated a vast number of medical images as seeing numerous patients. Images published in journals are publicly available for future researchers and radiologists to study as their interpretation is thoroughly explained. However, most medical images with high clinical value are saved and protected in hospital database for privacy and ethical reasons. Even though the internal radiologists trained on them can pass the ‘translational knowledge’ further, still, not all the data will be utilized due to the large volume. In such cases, those AI-powered medical imaging technology algorithms can learn those data and share the acquired knowledge with anyone who operates the machine, of course if with ethical permission. In this way, AI-powered medical imaging acts as a ‘somewhat experienced’ doctor, which will not only take good advantage of the historical invaluable medical imaging data but also train other radiologists with acquired experience. 

Implementing AI can also help reduce the occurence of medical errors like how it works for the autonomous driving system. 

Due to the complexity and the unpredicted ongoing progress of the diseases, there will be perception variability of one single medical image among observers. 

The incorporation of AI enables a ‘peer-reviewed’ process during the operation of the medical imaging tool. It allows comparing the newly generated images with all the historical clinical cases. 

Therefore, the observer variability within each doctor is highly reduced, as is backed up by previous clinical data. AI can also detect and identify imaging pattern changes that are not readily amenable to human identification, thus further reducing the chance of misdiagnosis. 

Public data showed that in 2016 there were estimated 7.54 billion medical visits for radiology in China but the number of radiologists was 158,000. There is continuous lack of available radiologists while the need for radiologists is stably increasing. The high workload pressure is another contributing factor to the misdiagnosis. 

AI can also take the tedious and repetitive tasks away from radiologists through performing tasks such as automatic protocol generation and smart reporting. Such intelligent processes can relieve the doctors’ pressure and reduce their workload. 

Currently in China, AI-assisted medical imaging technology has been applied to a variety of diseases including lung cancer, breast cancer and diabetic retinopthy. 

Some of the top players in China such as InferVision, Shukun and United Imaging Intelligence all have received more than one medical device class III certificate covering multiple indications including lung and breast from NMPA (National Medical Products Administration). Only with the medical device certificate granted from NMPA, can those medical devices be freely manufactured and traded legally. Up to now, more than 60 medical device class III certificates granted for AI-aided medical imaging by NMPA in China. 

In 2021, Keya, Airdoc and InferVision consecutively submitted the prospectus and aimed to be listed in HKSE.Airdoc finally went public on 5th November 2021, becoming the first share of AI medical imaging at an issue price of HKD 75.1 per share. The secondary market is still concerned about the competitiveness of companies’ core products. Challenges such as long-term losses and commercialization difficulties are still what most pre-IPO enterprises are facing. 

“What is happening at the back of those challenges now is that the level of the clinial demand is strong, but the level of purchase desire is far behind. Those ongoing projects in between the gap will certainly not generate any monetarized value in the short term. However, they are part of the R&D expenditure.” Lv Chenchong, the founder of Yizhun.AI told EqualOcean. “Taking the development of AI medical imaging industry as playing a puzzle game, we are now working on many of those single pieces towards the final work; looking at the long term, it certainly is a sustainable business.”

Switch to more sustainable adoption 

Here we roll out an ‘SPC’ model and integrate the industry expert views to make recommendations on how to facilitate a better commercialization strategy for the AI medical imaging industry from three dimensions: Strategy and shared value, Product and tech-driven and Customer and operation. 

In terms of strategy and shared value, the basic rule which applies to any other industry is that it is fundamental that the background and experience of the founder should have shared value with the relevant industry, in this case, artificial intelligence or medical imaging technology. The founder with artificial intelligence background can help set a more clear direction for product development by knowing well how to customize the products as a solution provider expert. The medical imaging background enables the founder to think more from the consumer side, who truly understands the pain points in the industry. Consequently, it will help launch more feasible and acceptable product strategy planning. Either background will help shape the company value and shift the product to be more market demand-driven. 

Sun Yuhui, the founder of Vistel, told EqualOcean, “I worked for Intel for many years and had been involved in enabling medical device projects. I figured out that the medical device is getting smaller and cheaper driven by Moore's law. Thus, more medical devices are affordable for entry level hospitals or even individuals. When those medical devices generate tons of data, there's no way for doctors to interpret the data and make it useful for the users because it is just too much. That's why AI plays a huge role here to make sense of the data. The whole world is getting smarter, and the medical world should not be legging behind. This is one of the reasons why I set up Vistel - to make the diagnosis and treatment process smarter and add value to doctors and patients.”

Stemming from the shared value, a feasible market strategy or expansion integrating the understanding of the technology, market needs and environmental regulations can be established. 

From the perspective of product and tech-drive side, the underlying technology of AI-aided medical imaging products needs to be capable of solving the core needs of doctors and hospitals. Moving from single-solution provider production to multiple-solutions provider production is one of the future trends for AI medical imaging companies. “Our hospital has been continuously pushing the implementation of AI-based medical imaging from one disease to multiple diseases and from a single organ to multiple organs. Except for diagnosis, we also use them to do therapy, and we recently are looking into other new projects.”Dr Hu Hongjie, Director of Department of Radiology in Run Run Shaw Hospital, Zhejiang University School of Medicine told EqualOcean. Companies also need actively carry out workshops to strengthen their connections with hospitals and reach a mutual understanding of their needs. 

Meanwhile, to survive in the fierce market, companies need to be clear about their product differentiation. Except for advancing the technology, the targeted group of users also requires a clear definition. For example, for county hospitals that have little access to advanced technology, it is necessary for the companies to hold product education sessions locally to well explain the products and make sure the product really plays its role.

When it comes to ‘Customer and operation’, EqualOcean firmly believes that globalization is the key to boosting China-based AI medical imaging companies’ reputation and value. China has a huge and strong talent pool of artificial intelligence. According to the latest Stanford University’s AI Index which assesses AI advancements worldwide across various metrics in research, development, and economy, China ranks among the top three countries for global AI vibrancy. On the other hand, China’s strong labor force has established a solid foundation for it to be the manufacturing hub. “The global need for intelligentization is huge and most of the current intelligent product suppliers are based in China.” Lv Chenchong, the founder of Yizhun.AI told Equalocean, “Speaking of the foreign market, it is definitely what Yizhun.AI’s ambition is going for. Now even though we have not officially started the international marketing project, we already received collaborational requests from the overseas, from which, you can see that China’s AI medical imaging technology is being naturally appealing to the overseas.”

Dr Hu Hongjie, Director of Department of Radiology in Run Run Shaw Hospital, Zhejiang University School of Medicine, also has promising expections about Chinese AI medical imaging products as he said, “I think China’s AI medical imaging products have high market prospects in the international market, especially about the backend maintenance services.” 

Conclusion

Sustainable business strategic planning is what AI-assisted medical imaging companies should be fed on in terms of the long-term development. Although the implementation of AI-based medical imaging tools has been seen in many hospitals in China, the continous improvements on the technical side are needed. On the basis of high-quality products and well-established backend services, the products need be capable of meeting the core needs and solving the pain points that radiologists are having. 

To cover the downside of the commercialization in the industry, in the short term, developing products targeting the purchase desire could be the best outlet out of the situation.